Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

CoRL 2026: 9–12 November 2026, AUSTIN

Humanoids 2026: 6–9 December 2026, SANTA CLARA, CA

Enjoy today’s videos!

A quadruped robot autonomously solves a giant Rubik’s Cube using only its feet.

[ ETH Zurich Robotics Systems Lab ]

Humans effortlessly react to the world; they adeptly return a tennis shot, catch a falling cup, step aside from a passing cart, or pass a box tossed during truck unloading. These everyday actions require anticipating a moving target and reflexively responding with the whole body within about a second. Such visual reactive whole-body control remains challenging for humanoids. We present Reflex, a three-stage framework that learns these capabilities stage by stage while connecting them through a compact representation of object dynamics.

[ Reflex ]

Thanks, Weikai!

I have long advocated for exactly this in the context of robotic laundry folding, or more broadly, other robot recognition and manipulation tasks. Why not make it easy when we can make it easy?

[ ACM SIGCHI ]

We first wrote about Parallel Systems in 2022. Nice to see them still rolling.

[ Parallel Systems ]

I’ll be honest with you: The end of this video is freaking me out a little bit for some reason.

This video showcases the whole-body loco-manipulation capabilities of KAIST Humanoid v0.9, developed at the DRCD Lab in collaboration with RAILAB, which developed the robot’s arms. Using onboard visual feedback, the humanoid demonstrates pushing objects weighing up to 35 kilograms and lifting objects up to 20 kg to arrange the letters “KAIST.”

[ KAIST DRCD Lab ]

When you don’t spend any money on sensors for your robots, you can afford much bigger wheels.

[ RaiLab Kaist ]

Those robots with wires have gotten somewhat weirder.

[ JSK Robotics Laboratory ]

The AthenaZero has two, three-fingered robot hands driven by tendons. This design allows the fingers to interdigitate to hold objects of different sizes, such as an egg or a hammer. Each finger is powered by a motor located in the robot’s torso to keep the hand light, transmitting force through flexible cables. This hand is both fast (closing in about 80 milliseconds) and strong. With the addition of highly accurate barometric pressure sensors, it can also be gentle.

[ Robotics and AI Institute ] via [ Science Robotics ]

A biohybrid robot combines living tissue with engineered materials—here, lab-grown skeletal muscle paired with a soft, 3D-printed scaffold that turns the muscle’s contraction into movement. What if you designed the living muscle and the soft scaffold together, instead of growing the muscle and then building a skeleton around it? In this work from the Soft Robotics Lab at ETH Zurich, we adapted a soft-body simulation framework to co-optimize both at once—and the resulting centimeter-scale bioactuators achieve up to 11x more range of motion than previous systems of comparable tissue volume. Strong enough to jump, swim, walk, and grasp.

[ ETH Zurich Soft Robotics Lab ]

Oh yes, this is definitely the use case that will scale humanoid robots.

[ Unitree ]

Here are a couple more teasers from the upcoming ACM Symposium on User Interface Software and Technology (UIST).

[ ACM Symposium on User Interface Software and Technology ]

In three hours, on a single GPU, SuperMimic surpasses the return of a state-of-the-art motion tracker baseline trained for seven days on 128 GPUs. It reduces GPU-hours by more than 1,000× and fails on only 25 of 129,514 benchmark motions, versus 179 failures for the baseline. After 9.8 hours on two GPUs, SuperMimic achieves zero failures.

[ Hybrid Robotics ]

This is getting out of hand! Now there are two of them!

Introducing the Grain Weevil dual robot attachment. Two robots linked into one larger machine, run by a single operator, doing the work of up to three robots.

[ Grain Weevil ]

Today [21 October 2026] at CES Unveiled Milan, Generative Bionics announced its plan to build Europe’s largest Physical AI Data Factory, with AMD as its AI infrastructure partner.

[ Generative Bionics ]

From robotic rovers exploring the rugged lunar landscape to lunar terrain vehicles carrying Artemis astronauts across the surface, mobility will play a crucial role in establishing an enduring human presence on the moon. During the early phases of Moon Base development, these vehicles will help NASA investigate the lunar South Pole region, conduct scientific research, test new technologies, and gather the information that will inform future missions. In this Moon Base update, get a closer look at some of the lunar terrain vehicles and robotic rovers helping to pave the way for humanity’s first lunar outpost.

[ NASA ]

I appreciate the designer of this single motor drone crediting previous research, which included some cool drones from UPenn that we wrote about in 2014.

[ Joshua Bird ]

Why choose a bipedal, human-form robot over specialized wheels, gantry arms, or quadruped platforms? Speaking at the Humanoids Summit Seoul 2026, Michael Perry from Persona AI tackles the single most debated question in the industry and delivers the definitive business and engineering justification for the humanoid form factor.

[ Humanoids Summit ]

Agility Robotics, the company I cofounded out of my robotics lab at Oregon State University in 2015, has just launched Digit 5. It’s designed to be the first humanoid robot that can safely and economically do useful work at scale, allowing us to bring tens of thousands of humanoid workers into the real world. Engineered to lift 23-kilogram bins onto high shelves and with a battery that can run for 90 minutes on a 9-minute charge, Digit 5’s goal is to handle heavy warehouse tasks shift after shift, filling the need for labor in jobs that are too physically stressful or too dull for humans to want to do.

Digit 5 is what Agility has been working toward for over a decade. But the robot is heavy and bulky, lacking the graceful natural motion of its predecessors, and it’s farther than I thought possible from the robot I dreamed of building when I started this company. To deliver a useful commercial product, Agility has taken a path that is very different from what I first imagined.

From ATRIAS to Digit

My dream has always been to build robots that combine locomotion and manipulation to move through the world and interact with it as people do. I seek a mastery of motion, achieved through devices designed to navigate the constraints of physics with elegance. When I started the Dynamic Robotics Laboratory at Oregon State in 2008, we began with a theory derived from decades of work in biomechanics: that a simple “spring-mass” model (a body supported by legs that act like springs) should be able to reproduce every known walking and running gait found in animals.

We put theory into practice by building the bipedal robot ATRIAS—short for “Assume The Robot Is A Sphere,” a joke among physicists about reducing a complex system to its simplest possible form. ATRIAS looked like a large microwave on stilts because it was only intended to demonstrate the physics of human walking and running. This was the first glimmer of achieving my dream; even though ATRIAS wasn’t designed to look like us, it still seemed natural, and even somehow alive, when it settled into a walking gait.

With the success of ATRIAS, the barriers to progress shifted from scientific to engineering and organizational, and a startup company seemed like the way to bridge the gap from a one-off research concept to an actual robotic product. I felt confident of two things: that I would be able to build robots that could go where people go, and that this capability would be incredibly useful. This combination of ambition and naivete is necessary for anyone to found a new company—how hard could it be?

Good advice for starting a company is to understand a customer problem, and then find a way to solve it using your unique capabilities. My cofounders, Damion Shelton and Mikhail Jones, and I did not do this. We relied on our belief that this technology would enable new possibilities and trusted ourselves to figure out the business as we went. So we simply focused on building our first product: the bipedal robot Cassie.

Cassie was the first dynamic bipedal robot that researchers could buy, able to maintain its balance through active, fluid motion rather than careful, rigid steps. It had ostrichlike legs, not because we were imitating ground-running birds, but as a result of engineering strictly for performance. It was not intended to be commercially useful, and had no torso, arms, or head. We sold Cassie robots to Harvard, Caltech, the University of Michigan, the University of California, Berkeley, and a handful of other research institutions, ultimately saturating the market for US $250,000 bipedal research robots after having sold a total of 10 of them.

ATRIAS, Cassie, and Digit 4 EVT [left to right] are on display at Agility Robotics’ RoboFab in Salem, Ore. The 70,000-square-foot facility has the capacity to produce more than 10,000 Digit 5 robots per year.Agility Robotics

Because Cassie’s hardware enabled research into the physics of walking and running, the legged-robot research frontier shifted to control software. Traditionally, engineers controlled legged robots with detailed mathematical models that determined how the machines should move. With Cassie, researchers pioneered the new approach of reinforcement learning for bipedal legged locomotion: A simulated robot learns through trial and error, and the resulting neural network controls the physical machine. This approach is what enables virtually all of the dynamic behaviors that you see on legged robots today, and Cassie looked alive as it recovered from stumbles while climbing stairs, ran 5 kilometers across the Oregon State campus, and broke the robot world record for a 100-meter dash.

Cassie got us closer to being able to move through the world with confidence and grace, but to make it useful, we needed manipulation. After exhaustively exploring different configurations, we found that an upright torso helped the robot balance, while two arms mounted near the top of the torso allowed it to pick up large objects and to catch itself when it fell. Digit 1, which added a torso and arms to the Cassie legs, was our first mobile manipulation robot, and another step on the path toward a product that could do useful work for commercial customers.

Why Digit Began With Warehouse Bins

As we explored the first applications for Digit, potential customers kept asking us: Why legs? And why a humanoid? These were tough questions for me to answer, because they exposed a core business reality: It didn’t matter to Agility Robotics if we solved a cool science problem in a beautiful way. What mattered to the business was solving real problems for customers.

After years of engagement with industry partners, we found one answer in warehouses and manufacturing facilities: moving bins. Bin handling is a clearly defined task where two arms are needed to lift large bins up high and legs are the best way to balance and move in narrow spaces. Because the task can vary widely from one deployment to another, it hasn’t been broadly addressed by single-purpose automation. Yet it was constrained enough to give Digit an economically viable first job while it developed a broader set of skills.

In 2019, Digit began demonstrating its capabilities to do real work. With Ford Motor Company, we showed how Digit 1 could carry packages from an autonomous vehicle to a front porch. We developed Digit 2 and 3 while working with a certain global package delivery company on moving and sorting cardboard boxes. Amazon tested Digit 4 on moving and stacking bins, and it became the world’s first commercially deployed humanoid at the logistics provider GXO. In 2024, Digit appeared on the cover of Time magazine.

Despite Digit’s apparent success, we began to run into conflicts between my dream and the needs of actual customers. We were no longer working with relatively patient university research labs. Agility has to sell robots to survive, and customers will only purchase robots that provide value. If bins weren’t moving, our customers were losing money. We spent a huge amount of time and effort keeping our first deployments operational, which arguably eliminated any profit we made. Yet this also kept our company learning and alive the way no lab demo ever could.

I had no idea how hard it would be, or that running a company would involve so many things that have nothing to do with solving technical problems: deployments across the country; shipping, service, and customer support; fundraising and investor relations; more fundraising; human resources and IT; developing fleet-management software; managing a supply chain; and even more fundraising. I had no experience with any of these things. My career had focused on solving the technical challenges of robots interacting with the world, and it turned out that was only a tiny piece of building a successful robotics company. But Agility was the only path I saw to continuing the technical work and having an impact.

Turning Digit Into a Product

I knew that Agility needed to focus more on robots as a product, but I didn’t know how to make that happen. Hiring the people we needed to run the company meant giving up an immense amount of control, but we had no other choice. We recruited a new executive team with the experience required to commercialize this new technology, and they quickly made changes that were both painful and necessary.

Digit versions 1, 2, and 3 [left to right] show the evolution of the robot from a dynamic walking platform into a useful mobile manipulator. Agility

We had intended the next revision of Digit, Digit 4 EVT (Engineering Validation and Test), to be a relatively minor upgrade. But Agility’s new leadership insisted that the robot was missing key features that were essential if Digit was going to become a real product. So Digit 4 EVT instead incorporated a series of major changes in a highly compressed timeline, such as larger batteries and system-wide upgrades for safety and regulatory compliance, which added enough mass to dramatically alter the robot’s appearance and capabilities.

While early Digits were headless, Agility Robotics was able to justify a head for Digit 4 to optimize antenna and camera placement, and it turned out to be surprisingly useful for human-robot interaction.Christie Hemm Klok

I could only watch as others brought their own talents to bear to make Digit into a successful product, in gut-wrenching ways that didn’t always align with my vision. There were some especially difficult times when I felt like there were compromises made that didn’t seem necessary, but making those decisions was no longer my sole responsibility. I had to accept that Digit had grown far beyond me, and look for ways of influencing decisions rather than directing them. In the moment, it was incredibly frustrating, as Digit 4 EVT took a few steps backward from my vision of how humanoids should look and move.

Agility is in our name because agility is core to the future of the company. I want Digit to be lithe and graceful. When people see Digit, I want them to like the robot, to understand what it can do at a glance, and to feel that it’s there to help. But with Digit 4 EVT, all of that was sacrificed to functionality, because of course it had to be—the customer just needs the robot to do the work. The design process became a ruthless triage, with aesthetics and elegance of movement being delayed over and over again in favor of brutal functionality. For an engineer whose mind lives more in the future than in the present, it has been hard for me to accept this trajectory. And yet, through Digit 4 EVT’s commercial deployments, we began to really understand specific applications and what they required. We mapped the first 10 use cases Digit would perform, and then began designing Digit 5 from scratch based on everything we had learned.

Digit 5

Digit 5 will be Agility’s first real step toward a general-purpose robot. It marks our transition from imagining what a future humanoid robot should be to designing a product around the realities of customer deployments. These realities include things like European CE regulatory compliance, North American UL electrical-safety certifications, and Federal Communications Commission limits for electromagnetic radiation—all of which felt tangential to my dream for Digit, but were absolute requirements.

Digit 5 is designed to be the first humanoid robot that can operate safely next to human coworkers while working 20 hours a day and lifting more than a human can.Christie Hemm Klok

Customer requirements helped us realize that safety in particular was one of the most critical barriers to scaling. In Digit 4’s commercial deployments, the robots had to work behind physical barriers, separated from humans at all times, to meet insurance and regulatory demands. This seemed silly to me, because I’ve never felt that our robots were dangerous, but we have to prove that they’re safe. And since keeping the robots in safety cages severely limited their usefulness, enabling Digit 5 to operate safely around people became a primary design goal. We spent two years on this challenge.

Digit 5 is designed to be the first humanoid robot that can operate safely next to human coworkers. It will rely on an array of sensors to locate nearby people in a manner that’s certifiably safe, and then take actions based on that information. As a person approaches, Digit 5 is programmed to stop what it’s doing, set down its payload, sit on the floor, and cut power to its motors before the person gets close enough to touch the robot. This conservative approach reduces the risk to nearby people in a verifiable way, and we believe it’s the only approach that makes sense as a first step toward working alongside humans.

Yet this strategy also feels like a massive slowdown on the path toward robots becoming part of everyday life. Enabling coexistence with humans has influenced almost every aspect of Digit 5’s design: the shape of the legs and the body, the safety-rated sensors within protective housings, the isolated computers dedicated to safety, and more. All of these features increase the robot’s complexity, power consumption, mass, and physical size.

Digit 5 is Agility’s largest, heaviest, strongest, and most capable humanoid robot worker.

The robot’s design was driven by safety, with integrated protection for the robot and for the humans it will work around.Agility Robotics

Weighing over 120 kg, Digit 5 is a huge robot. It needed to achieve the lifting limits recommended under the U.S. National Institute for Occupational Safety and Health guidelines for human workers, which include repeatedly lifting 23-kg bins and reaching shelves 1.75 meters high. Digit is designed to perform tasks like these for 20 hours a day—something no human can do. This requires motors with enough power and torque to lift the load, but also enough precision to delicately control the force Digit applies when it makes contact with the world. To power those motors, we needed a large battery and a dock for rapid autonomous charging. The dock will deliver 240 amperes at 48 volts, which is enough current to make its thick charging cables visibly wiggle under the induced magnetic forces, while powerful fans blow air through cooling channels in the robot’s torso to keep the battery from overheating. The battery and cooling system takes up about 75 percent of Digit’s torso, but it will allow the robot to operate for 90 minutes and fully recharge during a 9-minute break, enabling near-continuous work.

All of this extra bulk means that Digit 5 can no longer run or jump the way previous generations could, and the svelte form I imagined has become physically impossible. In some ways, Digit 5 feels more like a forklift than the graceful humanoid I set out to build. But looking at the situation rationally, I believe this trade-off is necessary for any company that claims its humanoid can do this kind of work today: If you see a robot that looks like a human working in a factory or warehouse, it either can’t lift the required weight or can’t make it through the whole shift. And if it’s not covered with sensors, it’s likely not safe to stand next to.

The most visible change with Digit 5 is that humanlike legs replace the ostrichlike legs of our earlier versions. While the “backward” leg configuration was optimized for walking and running, we discovered that it created problems with squatting because the ankle will strike the ground. Digit 5 will spend a lot of time lifting heavy objects from near the floor, so we used a humanoid leg configuration with knee and hip motors that both contribute torque to help the robot stand up again.

When Digit detects an approaching human, the robot will lower itself to the floor and cut power to its motors. Agility believes this is the only near-term approach to verifiably reduce risk when the robot is working near people.Agility Robotics

Other companies building humanoid robots have copied the human form without ever understanding its purpose. It’s very frustrating that, to an outside observer, it now looks like we discarded Agility’s distinctive leg design and simply did what everyone else was doing. Yet I know that we arrived at this shape based purely on functional engineering choices. Making a robot look like a human is the easy part—without the function to match, and the understanding of the physics underlying that function, the whole point of the human form may be lost.

With the power, battery life, and safety features that we are designing into Digit 5, we will have a truly useful robot. But it also compromises on some things that are core to my vision and mission. I understand why we made the choices we did and can justify them all, but together they leave me unsatisfied. It’s the curse of the engineer.

What Comes After Digit 5

I believe Digit 5 will be the scaling moment for Agility Robotics, the inflection point where we can begin deploying robots in large numbers. While the robot will continue carrying bins, our road map includes many more tasks, from picking individual items and tending machines, to preparing parts for assembly and shipping, and eventually doing assembly work itself. Its interchangeable tool interfaces and modular design will enable us to gradually upgrade its subsystems. It will be a platform for improvement, a source of data for future behavior training, and a viable product all at the same time.

Hurst stands between the previous version of Digit, Digit 4 EVT [left], and Digit 5.Christie Hemm Klok

But I cannot love Digit 5. It’s the largest, heaviest, most expensive robot Agility will ever build, full of design compromises that enable it to be exactly the robot it needs to be right now rather than the robot I dreamed of. In the beginning, I envisioned steady progress on agility and dexterity, with each version becoming more dynamically capable. Instead, developing Digit 5 has required large detours to incorporate requirements that seemed unrelated to my original vision. However, I now recognize that tackling these challenges was always necessary. Customers and product requirements didn’t ruin my vision; they revealed what my vision was missing. This is how robots will achieve commercial success and make a real-world impact, which will bring me closer to my scientific dream as well.

We won’t have a humanoid that matches the capabilities of humans until it achieves that mastery of motion that inspired me to found Agility over a decade ago. I saw glimmers of it with the spring-mass locomotion of ATRIAS, and more glimmers with Cassie’s running gait.

I know that such mastery is physically possible with the tools and hardware that we have right now, and despite all of its compromises, Digit 5 will give us a practical foundation for pursuing that dream. I want every future version of Digit to be lighter and more graceful, to interact more naturally with humans, and to be better at manipulating objects and moving through the world, enabling robots to enter aspects of our lives that they haven’t been able to help with before. By the time I’m old enough to need assistance at home, I expect Digit will be up to the task.

With Digit 5, Agility has built a robot that will be useful in our world. Now I’m getting back to work to make it beautiful, too.

This article appears in the November 2026 print issue.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

CoRL 2026: 9–12 November 2026, AUSTIN

Humanoids 2026: 6–9 December 2026, SANTA CLARA, CA

Enjoy today’s videos!

When one robot loves another robot very much, and they’re connected by a wire, they can get together and make a brand new robot! SFW.

[ JSK Lab ]

What happens when Microducks meet? They start to sing together. Each Microduck syncs with its neighbors over Bluetooth, then they compose a song on the fly, in harmony and in perfect time. There’s no score and no conductor. Two, three, or four ducks are enough for a choir to break out.

If Pollen isn’t selling Microducks in four packs for this exact reason, WHAT ARE THEY EVEN DOING!?

[ Pollen ]

In an internal test, Google DeepMind robotics researchers instructed Gemini 4 Argon to watch a video of a human playing cornhole. Using its video understanding and advanced reasoning capabilities, Gemini 4 built a virtual simulation in MuJoCo and accurately calculated the physics and whole-body control required for Apptronik’s Apollo 2 humanoid to sink seven consecutive shots in the real world. This Video-to-Sim-to-Real pipeline compresses weeks of manual modeling into a few hours, transforming ordinary videos into an infinite training ground for physical AI.

[ Google DeepMind ]

As of 11 October 2026, NASA’s Perseverance rover has driven 28.10 miles (45.22 kilometers), becoming the new off-world record-holder for total distance driven by any vehicle on the surface of another world. The previous record of 28.06 miles (45.16 kilometers) was held by NASA’s Opportunity rover. Perseverance reached that mark on the 2,006th Martian day, or sol, of its mission, covering in five and a half years a distance that took Opportunity almost 15.

[ JPL ]

Let’s not argue about whether these are robots or not and just agree that they’re cool.

[ OATs ]

If one cute little robot can clean up one block, how many blocks can 40 cute little robots clean up? Science has the answers!

[ Paper ]

Thanks, Keisuke!

Before Rosalind Franklin can explore Mars, it has to get off its lander. Engineers are rehearsing that first journey on Earth, using a full-size rover model and Mars-like terrain. The test recreates some of the challenges Rosalind Franklin will face on the Red Planet, from reduced gravity and a 20-degree incline to uneven ground and obstacles beneath its ramps.

[ European Space Agency ]

From custom part to running robot, the Atlas shop floor is built on precision, rapid assembly, and real-time collaboration. By machining parts in-house and streamlining assembly and repair under one roof, we are driving the manufacturing velocity needed to turn complex designs into walking hardware at scale.

[ Boston Dynamics ]

What can a quadruped robot do beyond the lab? Deep in China’s Ailao Mountains, DEEP Robotics Lynx M20S joined a team of researchers on a biodiversity field expedition—navigating rugged terrain, scouting routes ahead, carrying research equipment, and capturing wildlife in areas that are difficult to access on foot.

[ DEEP Robotics ]

The Kamino simulation software is a GPU-accelerated physics solver designed for accurate simulation of complex mechanical systems with kinematic loops, such as parallel manipulators, planar and spatial linkages, and multi-limbed robots with coupled joints, without requiring simplified approximations.

[ Disney Research ]

The original Kawasaki UNIMATE robot as featured in (remember this show?) The MacNeil/Lehrer Report in 1982.

[ Kawasaki Robotics ]

WildOS enables autonomous robots to perform long-range, open-vocabulary object search in previously unmapped and unstructured outdoor environments by unifying geometric safety with semantic visual reasoning. Through a vision-scored navigation graph and beyond-horizon target localization, the system demonstrates more efficient and robust exploration than conventional geometric or purely vision-based approaches in extensive field experiments.

[ ETH Zurich Robotics Systems Lab ]

Who needs an AI-generated person when you can have an actual robot that looks exactly like an AI-generated person instead?

[ UBTECH Robotics ]

The recovery behaviors here are impressive, but the robot is far slower than a human would be, suggesting that the value it generates is primarily in novelty. Which is fine for now, but novelty wears off, and then what?

[ Sharpa ]

Here are a couple of good panels from ICRA 2026 in Vienna, including one on “Impact and Return on Investment on Humanoid Robots” and a second on “Advancing Sustainability in Robotics.”

[ ICRA 2026 ]

On 11 October, the Perseverance rover completed a drive that brought its total distance traveled on Mars to 28.10 miles (45.22 kilometers). This is a new record for Mars rovers, enabled in large part by Perseverance’s ability to drive with much more autonomy than its predecessors. Over the past five years, Perseverance’s autonomy software has been steadily upgraded, and to learn more about how the record was broken and what the future holds for the rover, we spoke with Mark Maimone, the Perseverance Robotic Operations Deputy Team Chief at NASA’s Jet Propulsion Laboratory, who designed and developed the self-driving software for Mars rovers Spirit, Opportunity, Curiosity, and Perseverance.

Before we talk about Perseverance, let’s pour one out for the previous record holder, Opportunity, which landed on Mars in January 2004 for a three-month mission and ended its epic journey 15 years later in February 2019. This NASA video compresses Opportunity’s entire mission into four minutes and gives a poignant sense of the incredible distance that this little rover traveled over its 15-year life on Mars, doing science as it went:

NASA Jet Propulsion Laboratory/YouTube

Opportunity drove 45.16 kilometers until a planetary-scale dust storm stopped it in its tracks, appropriately enough, in Perseverance Valley.

On 18 February 2021, Perseverance joined NASA’s Curiosity rover on the red planet, carrying the Ingenuity helicopter tucked under its belly. Curiosity and Perseverance both ditched Opportunity’s solar panels for a radioisotope thermoelectric generator. Curiosity has been active on Mars since 2012.

Despite Curiosity’s nearly 10-year head start, Perseverance is the rover to first break Opportunity’s record. There are several reasons for Perseverance overtaking Opportunity—ruggedized wheels for one—but the biggest contributor is that Perseverance was built for autonomy.

Maimone shares how JPL added enough autonomy for its Mars rover to drive itself from 225 million km away without crashing.

This animation shows Perseverance’s point of view during a drive of 807 feet (246 meters) along the rim of Jezero Crater on 10 December 2025, the 1,709th Martian day, or sol, of the mission. Captured over 2 hours and 35 minutes, 53 Navigation Camera (Navcam) image pairs were combined with rover data on orientation, wheel speed, and steering angle, as well as data from Perseverance’s Inertial Measurement Unit, and placed into a 3D virtual environment. The result is this reconstruction with virtual frames inserted about every 4 inches (0.1 meters) of drive progress. JPL-Caltech/NASA

Let’s talk about rover autonomy on Mars. Even before Perseverance, rovers were driving themselves, right?

Mark Maimone: Yes! Sojourner [in 1997] had an early kind of autonomous behavior—for sensing, it used four laser stripes projected onto the surface, which it used to measure the height of 20 points in the terrain in front of it. It could steer around big obstacles, or even thread the needle between two things. It was a real autonomy system, which is very impressive considering its CPU was less than one megahertz.

With Spirit and Opportunity, I developed the autonomy for those rovers. Every time they drove, they would use 256 x 256 pixel stereo pairs to map the geometry of the terrain and find a safe route forward. Spirit actually ended up driving itself 23 percent of the time, which I was frankly shocked by, since it drove much more slowly in that mode. But using autonomy helped Spirit travel 50 percent faster than it otherwise could have. Opportunity and Curiosity both only used autonomous driving about 5 percent of the time. Perseverance is the first Mars rover that drives itself 90 percent of the time, with its enhanced autonomy system that includes a second processor with an FPGA to speed up extra image processing operations.

What was it like to let Perseverance drive itself for the first time?

Maimone: Perseverance is the culmination of decades of work on Mars rover autonomy, so seeing Perseverance use its AutoNav system to drive itself basically at full speed, up to 120 meters an hour, was incredible.

These two images taken by Perseverance’s hazard avoidance and navigation cameras show (at left) the very first AutoNav drive on Mars where the rover autonomously stopped before running into a rock pile, and (at right) the second AutoNav drive where the rover successfully steered around a rock.JPL-Caltech/NASA

Where are the humans in this process?

Maimone: The plan of where to go is specified by the science team, and then on the engineering side, the strategic route planning team develops a path to get us to that goal by guessing, based on lower-resolution maps from orbit, what the terrain is going to look like for the rover. Each day, a tactical operations team combines the strategic overall route with that day’s science goals to pick the actual route the rover will follow.

We want humans to keep doing long-range planning because that avoids wasting time. For example, we’ve learned that if we see sandy areas with intersecting waves or crests—not just ripples blown in one direction, but cross-ripples going in multiple directions—the rover tends to sink into terrain like that, so we tell the rover to keep out of those areas. The onboard autonomy only measures terrain shape, not terrain type, so it benefits from those location-based annotations.

We now have years of experience correlating images taken from orbit with images taken by the rover, and understanding how different kinds of terrain affect the drive. With the current orbiters at Mars, we sometimes get 25 centimeters resolution on orbital images. That’s still not quite enough to see all obstacles; there could be a big rock that doesn’t really show up on a scale of more than one or two pixels, making it very hard to notice. That is why you want to have autonomy on the rover, but the greater help we can give it at a high level, the faster we can reach our goals and the fewer potential delays we’ll encounter.

This orbital view shows a region of complex Martian terrain that Perseverance could not navigate autonomously. The blue line shows the path that human planners estimated, while the white line shows the path taken by the rover. The inset image, from Perseverance’s hazard avoidance camera, shows the rocky, sandy, and sloped terrain.Source images: Olivier Toupet, Masahiro Ono, et al.; JPL-Caltech/University of Arizona/NASA

What happens when the autonomy system on Perseverance fails?

Maimone: We ended up not using the autonomy at all for about two months in more complex terrain while we tuned things and wrote new software to deal with that new terrain we hadn’t encountered before. Some of those changes were relatively quick, and some required months of study. For example, one parameter is the maximum amount of slope that the robot is allowed to climb, because on steep slopes, there is a risk of the rover slipping sideways.

We landed inside Jezero Crater, which was very flat, and so for the first three years, most of our driving was at less than 14 degrees of slope. When we finally decided to leave the crater, we began to encounter other terrain at higher tilts with greater rock density, and the experience we gained allowed us to re-tune the parameters to allow the rover to traverse those more difficult terrains, with slopes of up to 24 degrees.

Rover autonomy in the age of AI

Late last year, you let Anthropic’s Claude AI plan a 400-meter drive for Perseverance. How did that go?

Maimone: We had the AI look at the orbital images and pick out a path for us, which became the starting point of the day’s rover planning operations. Typically, that starting path comes from the human rover planner team, but in this case, it was generated by AI. Humans were still in the loop; the team reviewed it and made a couple of adjustments. AI gave us a solid starting point that allowed us to make rapid progress, because it’s much easier to validate a path than to pick one from scratch.

NASA’s Perseverance Mars rover took this selfie on July 23, 2024. The selfie is composed of 62 images taken by the WATSON (Wide Angle Topographic Sensor for Operations and eNgineering) camera on the end of the rover’s robotic arm.JPL-Caltech/MSSS/NASA

Do you think humans will remain a necessary part of rover operations, or do you see the next generation of Mars rovers operating for weeks at a time without human assistance?

Maimone: One of the things that has held us back from driving very long distances autonomously is that as we drive, the rover accumulates uncertainty in its position. When the uncertainty grows, it means we have to stay farther away from areas that human rover planners have identified as being potentially dangerous, called “keep-out zones.” In some cases, the uncertainty gets high enough that the rover can’t plan a safe path, because the keep-out zones (which grow with the rover’s uncertainty) get too big.

We now have a new capability that we’ve added to Perseverance called Mars Global Localization. With this new technology that can shrink our uncertainty, we theoretically have the ability to go as long as we have resources to keep driving safely based on our maps and keep-out zones.

But to truly run a mission for weeks, you need to not only drive long distances, but also execute science operations autonomously. We have people here also looking at automating science operations by doing image processing onboard the rover to get a better understanding of what’s out there and where to focus the instruments’ attention. So I do think we are getting to the point where we can support long-period autonomous operations. It has been a slow evolution, but we have great capabilities now and a good understanding of what’s needed to make it work well.

As someone who has been involved with all of these Mars rovers—including one that got stuck—do you fundamentally trust Perseverance to operate autonomously?

Maimone: I do. We built the software, tested it, ran millions of simulations, and did lots of work in the field in many different test environments. It hasn’t steered us wrong yet. We’re looking forward to setting even bigger records.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

CoRL 2026: 9–12 November 2026, AUSTIN

Humanoids 2026: 6–9 December 2026, SANTA CLARA, CA

Enjoy today’s videos!

It’s possible that I wasn’t the only one who thought that Figure’s recent stunt was somewhat ridiculous.

Devanthro is, in fact, selling these plastic balls for $1,000 each, and they’ll arrive in time for Christmas.

[ Devanthro ]

Not that anyone uses ringtones anymore, but if we did, this would be absolutely everywhere, and I’d be loving it.

[ Pollen ]

Let’s not get too excited, but this threadlike linear motor is pretty sweet.

[ Ecole Polytechnique Fédérale de Lausanne LMTS ]

What is this robot for? It’s for doing exactly what you see here, that’s what it’s for.

[ RAI Institute ]

Humanoid robots still struggle to plan contact-rich whole-body manipulation from egocentric RGB and proprioception. Workhorse learns such manipulation from robot-free human demonstrations. On a real Unitree G1, Workhorse sorts boxes with its hands and a kick, catches a thrown box, and topples and climbs a suitcase.

Yeah that happens to me too when I have a communication failure.

[ Hybrid Robotics ]

A car wash that models your vehicle, then cleans it with robot arms? Interesting idea.

The recent reviews from this location suggest that the robots aren’t quite up to snuff yet, but it really seems like there’s potential here.

[ KUKA ]

So this is not the worst idea I’ve ever seen for what to do with humanoid robots.

[ DEEP Robotics ]

Some lovely manipulation with a deceptively simple gripper.

[ Generalist ]

I am genuinely embarrassed to say that this robot dances better than I do.

[ General Robotics Lab ]

MIT researchers developed SANDO, a new navigation system that helps robots safely navigate unfamiliar environments while avoiding moving obstacles. The system can rapidly plan and adjust a collision-free path using onboard sensors and computing, with safety backed by mathematical guarantees.

[ MIT News ]

Can you spot the part of this video that I’m side-eyeing?

Yep, it’s where the middle clip starts after the hand already has a good grip on that peelable strip.

[ Sanctuary AI ]

Inspired by feather stars, the NC State researchers report a soft underwater swimming robot capable of achieving three distinct swimming modes—jellyfishlike vertical swimming, fishlike directional steering, and rotorlike 3D reorientation—using only two pneumatic inputs.

[ NC State University ]

Here’s a series of mini-keynotes from ICRA earlier this year on manipulation, humanoids, and embodied design.

[ ICRA 2026 ]

A hand attached to the end of an arm is a given for most humans, but what if robots didn’t have to play by the same rules as we do? Researchers from ETH Zurich have turned an off-the-shelf robotic hand into a completely self-contained robot that walks on its fingertips while using those same fingers to manipulate its environment.

Besides making the perfect Halloween accessory for anyone planning to dress up as a member of the Addams Family—who were famously accompanied by a disembodied walking hand called Thing—this robotic hand could make it possible to operate controls and handle objects in places a traditional robot arm can’t reach, the researchers say.

Soft Robotics Lab, ETH Zurich/YouTube

Amirhossein Kazemipour, a Ph.D. candidate in the Soft Robotics Lab at ETH Zurich who led the project, believes this hand could prove useful in contexts like fiddly maintenance tasks in industrial settings, or in search and rescue, where a robot could detach its hand to explore confined spaces too tight for the rest of its body. More broadly, he says, the work is a step toward a new approach to robotics in which body parts are not confined to a single role.

“I wanted to see how much a robotic hand could do without the rest of the robot,” Kazemipour told IEEE Spectrum. “Instead of having this traditional view of robotic parts, what if we could use those parts to do other tasks that they are not designed to do?”

This is not the first attempt to get a robot hand to walk, but previous work has typically relied on specially designed hands. The ETH team instead decided to use off-the-shelf hardware from Wuji Technology, augmented with an 80-gram backpack containing a battery, an inertial measurement unit (IMU), and a Raspberry Pi Zero. Kazemipour says they decided to go with a commercial robot hand because they didn’t want to compromise on the hand’s manipulation capabilities. This introduced significant challenges, he adds, because a hand’s shape is optimized for grasping rather than crawling.

How to Walk With Fingers

To teach the hand to walk, the researchers trained a model in simulation using reinforcement learning—an approach in which models learn skills through many rounds of trial and error, receiving rewards for getting closer to the desired behavior along with penalties when they diverge from it. But Kazemipour says the unusual geometry of a hand meant they couldn’t just reuse methods used to train other kinds of legged robots. Standard quadrupeds or bipeds tend to be symmetrical on their left and right sides, which simplifies walking and balance, he says. Hands, in contrast, are not at all symmetrical, with fingers of different lengths and an opposable thumb.

In addition to rewards for moving at the right speed and in the right direction and penalties for things like wobbling or jerky movement, the team introduced a novel way to keep the hand in a crawling stance during the simulated training. To account for unequal digit length, they gave each finger its own target position relative to the palm, based on where it naturally sits when in the crawling posture. They then penalized the fingers for straying too far from that reference point, creating what they describe as “virtual springs” that pull the fingers back to the crawling stance. Crucially, the penalties for fingers moving forward and backward are significantly less than those for moving side to side, so taking steps in that way costs the hand very little compared with the rewards it receives for moving.

The team also trained separate models for skills like righting the hand when it falls over, pressing keys on a keyboard, and pushing objects to a target. All four models fit on the onboard computer, says Kazemipour, and the controller can switch between them as necessary. In real-world tests, the hand crawled across 14 different surfaces including smooth floors, metal grates, grass, and gravel. Other experiments confirmed that it could make both right and left turns, and hit an average speed of 9 centimeters per second. The hand was able to right itself in 21 of 25 trials. When given a video feed from an overhead camera, it even managed to autonomously push a 41-gram cube to targets up to 40 cm away 15 times in a row.

“What interests me most about this work is that the same fingers can move the hand, support its weight, and interact with its surroundings,” says Masahiko Inami, a professor at the University of Tokyo whose group has developed hand-shaped walking robots. He agrees that a mobile hand could be useful for reaching confined spaces and operating controls that an entire arm could not easily access, though he says practical deployment would require onboard perception and more robust navigation. Hideki Shimobayashi, a Ph.D. student in Inami’s lab, also points out that walking puts very different loads on a hand compared to grasping, so durability may be a challenge.

However, Matei Ciocarlie, an associate professor of mechanical engineering at Columbia University, says it’s a “cool result” that suggests robot hands don’t have to be restricted to the same capabilities as their human counterparts. “One can imagine an entirely new class of mobile manipulators that are simultaneously dexterous, able to traverse complex terrain, and able to manipulate payloads comparable in size to themselves,” he says.

Kazemipour sees his work as part of a longer-term vision for robots made up of parts that are not tied to a single job or even a single body plan, something he calls “autonomous modular embodiments.” “Robotic body parts should not be permanently assigned to a single function,” he says. “They can dynamically transition between roles, between being components of one embodiment system and autonomous agents, according to the task.”

This White Paper gives robotics researchers and engineers an overview of a new large-scale motion capture dataset built to close the data gap limiting humanoid robot learning. It also shows how policies trained on the dataset transfer to a real humanoid robot.

What you will learn about:

• Why humanoid robot learning, a central problem in embodied AI and Physical AI, needs data that internet video and existing motion capture datasets cannot provide.
• How FrameNet, a linguistic framework for human action, can guide motion capture collection to systematically cover a broad range of whole-body motion.
• Why synchronized object trajectories and meshes make human-object interaction data useful for teaching robots real-world tasks such as carrying, pushing, and pulling.
• How reinforcement learning policies trained on this motion capture data improve with scale, and how sim-to-real transfer carries them onto a physical humanoid robot.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

CoRL 2026: 9–12 November 2026, AUSTIN

Humanoids 2026: 6–9 December 2026, SANTA CLARA, CA

Enjoy today’s videos!

We’ve developed ALBATROSS (Airborne Lander with Buoyant AuToROtating Sailing Sensor), a hybrid aerial-marine robot. It can be released from the air, autorotate down to the water without a parachute, passively self-right after landing, and then reuse the same rigid wings as sails for autonomous wind-powered navigation.The key idea is to make the physical structures do as much of the work as possible. Rather than carrying separate systems for aerial descent, landing, and marine propulsion, ALBATROSS uses the same wings for both autorotation and sailing, with minimal actuation. Ultimately, we’re interested in whether platforms like this could combine rapid aerial deployment with persistent, energy-efficient marine sensing.

[ Singapore University of Technology and Design ] paper via [ Science Robotics ]

Thanks, Shane!

I’m just getting so burnt out on the humanoid publicity stunts, you know?

[ Figure ]

Sharpa unveiled three major proprietary new products at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS): D01, the first integrated tactile-sensing robot built for dexterous manipulation; W02, a next-generation fully tactile, ultracompact, lightweight dexterous hand; and AE01, a high-fidelity haptic exoskeleton data glove.

[ Sharpa ]

Why will Boston Dynamics win the humanoid robotics race? We’ve already commercialized autonomous mobile robots, creating markets with Spot and Stretch. Now we’re doing it again with Atlas: learning in real environments, designing for manufacturing at scale, and introducing a new era of physical intelligence.

[ Boston Dynamics ]

This work presents a lightweight, compliant footpad that enhances the terrain adaptability of a hopping robot. The design combines a planar foot supported by a compliant spherical joint for self‑alignment with embedded spines to improve traction on uneven or slippery surfaces.

[ CLIMB Lab, University of Toronto ]

Heterogeneous robot teams distribute complementary capabilities across specialized agents, but their physical roles and capacities typically remain fixed throughout a mission. We present HARP, a Heterogeneous Aerial Robotic modules Platform, in which independently deployable aerial robots physically reconfigure to compose their capabilities for field operations.

[ General Robotics Lab ]

Some very impressive mobility from ANYmal.

[ ETH Zurich RSL ]

Daniel writes, “I’m a high school student in Canada, and I built a self-constructing robot arm that assembles itself. It’s designed to be low-cost, with disaster response in mind: Small robots can squeeze through gaps in rubble but are too weak to move things, while big arms are strong but can’t fit through. My arm is split into three wheeled modules that travel separately, then dock together using a spring-lock mechanism.”

[ Daniel Zhu ]

Autonomous highway driving not impressive enough for you? Try this.

[ XPENG ]

The IT Imperial team has connected MaleCNS—a published connectome of a male fruit fly (Drosophila melanogaster)—to a physical Unitree G1 humanoid robot. The system creates a closed loop: camera → fruit-fly neural model → motor decoding → physical robot movement. The robot receives visual input through its stereo camera. The images are transformed into a representation adapted to the fly’s visual system, processed by the computational model based on the reconstructed nervous system, and then converted into movement commands for the robot.

[ IT Imperial ]

Thanks, Anna!

My guess is that this is far easier than it looks or far harder than it looks, and I’m honestly not sure which.

[ Paper ]

I cannot believe that this is still a thing, but good on them!

[ Tevel ]

In Part 1 of Dexterity Explained, we break down what manual dexterity is, why it matters, and how robotic hands and physical AI work together to bring greater skill, precision, and control to physical tasks.

[ Sanctuary AI ]

XPENG has a lot of money to build a lot of these robots. They do look very human. I’m not sure what they’re going to do, though.

[ XPENG ]

That’s a lot of robots! Tell me what they’re all going to do!

[ UBTECH ]

When floods swept away roads and bridges in Nepal, local drone pilots stepped in to help map the damage, support search teams, and deliver essential supplies to communities cut off from help. In this episode of Ground Truth with Flying Labs, host Leka Tingitana speaks with Uttam Pudasaini, an advisor to Nepal Flying Labs, and Raj Bikram Maharjan, cofounder of Airlift Technology, about the drone community’s response to the devastating floods of 26 August 2026.

[ WeRobotics ]

Today, Boston Dynamics is announcing a redesigned hand for its Atlas robot. The previous generation could do some crazy superhuman things with its three fingers, but it was never intended for mass production in the tens or hundreds of thousands. Switching from research hardware to a scalable product makes previously ignorable questions suddenly paramount: How can you make a hand that does everything you need it to do but is also rugged and reliable and cost-effective? It’s questions like these that drove the design of the new Atlas hand, and while it doesn’t look much like a human hand, it might be a hand that makes it out into the real world.

- YouTube www.youtube.com

Compared to the almost eerily humanlike hands designed by a litany of other humanoid robotics companies, Atlas’s new hand does look awfully clunky. We’ve certainly seen some of those robots with much more humanlike hands perform dexterous manipulation tasks that are, speaking as someone who has personally witnessed robots haplessly fumble their way through the world for nearly two decades now, nothing short of incredible. And it’s not just what these hands can do—it’s how they look while doing it, as slim, graceful mechanical copies of our own hands. They seem like science fiction, and to some extent, that’s exactly what they are.

“Hands are a ruthless design trade-off,” Alberto Rodriguez, director of robot behavior at Boston Dynamics, tells IEEE Spectrum. “There’s no way around it, you’re always giving up on something. And many of the current designs are giving up on things like reliability and manufacturability.” This is what invariably happens when you try and cram everything necessary to make a robotic hand function within the form factor of a human hand: It’s going to be either very fragile or very expensive, and likely both at the same time. The unfortunate reality is that for now, and likely for the immediate future, all that these sorts of hands are good for are research and demos.

With Atlas’s new hand, Boston Dynamics is making the design trade-offs necessary to achieve a hand that, according to the press release, is capable of a vast array of useful tasks, including using tools; is strong and rugged; can be cleanly simulated; can be mass-manufactured and repaired at reliably low cost; and is close enough in form factor to a human hand that human demonstrations can be used to train it.

You’ll notice that “looks cosmetically like a human hand” is nowhere on that list, and this new hand doesn’t. Rodriguez explains that getting rid of the pinky was a straightforward decision, agreed on after his team spent a day with their pinkies taped to their ring fingers. But why even add a fourth finger, when the previous generation of Atlas’ hand seemed to do well with only three? It’s not just the extra finger, but a complete redesign that almost doubled the degrees of freedom between the old hand and the new, going from 7 to 13.

Boston Dynamics

“We also added the ability for the fingers to splay open because we knew that it was important for one finger to be able to move against another, and we also knew that we needed to be able to hold the handle of a tool with a trigger,” Rodriguez says, which includes many common tools like drills or welding torches. “We’re convinced that from a product perspective, this hand should be able to do everything we need it to do for the foreseeable horizon.” And that ability to finger splay goes beyond what human hands are capable of. According to Rodriguez, “With reinforcement learning, we can discover uses for these superhuman extra motions and exploit them.”

Reliability, manufacturability, and cost are all closely related. By reducing the degrees of freedom relative to other more anthropomorphic hand designs down to just 13, Boston Dynamics can use fewer, larger, and more powerful actuators. The actuators are embedded directly into the joints themselves in a direct-drive configuration, with a transmission that allows the motors to be back-driven and react to force and contact. Each actuator pack is a single unit that can be easily replaced, and there are no delicate tendons or cables across joints to stretch and break. “What we’re working on right now is figuring out what needs to change in the fine details of our design if we want to make 100,000 of these hands a year,” Rodriguez says.

Not every humanoid company will make the trade-offs that Boston Dynamics has done here, and there’s still plenty of room for creativity when it comes to hand design. But if humanoids with hands are going to scale, some difficult decisions will have to be made, and it’ll take some compromise to deliver on the promises that the humanoid industry has committed to.

Building useful robots starts with understanding the people who use them. For Charlie Kemp, cofounder and chief technology officer of Hello Robot, that means developing assistive robots that can help people with everyday tasks and support greater independence.

In this Robots Guide profile, Kemp shares his path from studying artificial intelligence at MIT to building Stretch, explains how working with people with disabilities has shaped his approach, and offers advice for aspiring roboticists. Read the full profile on IEEE’s Robots Guide.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN, TEXAS

Humanoids 2026: 6–9 December 2026, SANTA CLARA, CALIF.

Enjoy today’s videos!

Life’s better with a little Goose.

[ The Robot Works ]

Thanks, Dave!

Skydio has introduced the F10, a charmingly lopsided fixed-wing drone, that uses a robot arm for autonomous launch and capture.

Unfortunately, based on the frequency of that flashing LED on the drone, this video appears to have sped up the capture (and likely the launch) by a substantial amount.

[ Skydio ]

I want a bunch of these little guys. I don’t know what I’d do with them. But I want them.

[ Freeform Robotics ]

Walking robots are everywhere now…. But 30 years ago, they were brand new. The first ever robot that could walk by itself was the Honda P2. The P2 featured motors and hydraulics in the hips, knees, and ankles to walk around, and weighed over 300 pounds.

[ IEEE Spectrum ]

Precision biomimetic dexterous hand, priced from US $6,500.

[ Unitree ]

What if all it took to train a drone to do acrobatics was to tell it whether you like the way it flies?

[ UZH Robotics and Perception Group ]

Why build a humanoid hand when a robot hand can do stuff like this?

[ General Robotics Lab ]

Diane Heinle, a mechanical engineer at Boston Dynamics, answers questions about endurance testing for Stretch, how her team simulates years of wear on robot parts over a brief time in the lab, and why there are bird heads on them.

[ Boston Dynamics ]

ULOHA brings learning from demonstrations to two robot arms working underwater. We developed custom leader–follower hardware and software extending LeRobot to connect teleoperation, demonstration collection, and autonomous manipulation. The video shows the robots lifting objects together, passing objects between arms, opening a container, and catching a sponge as it rises through the water. We also explore how bubbles and action-execution timing affect learned behaviors.

[ ULOHA ]

Thanks, Masato!

Usually, ballbots rely on an internal pendulum, but here’s a different approach.

[ General Robotics Lab ]

From long-horizon manipulation to tactile-guided actions, this video highlights selected collaborations that explore what Astribot robots can do.

[ Astribot ]

This is one of those tasks that looks a little boring but is actually both very tricky and very useful.

[ HiPeR Lab ]

That tiny little excavator playing in flour is adorbs.

[ ETH Zurich Robotic Systems Lab ]

How do you fit a humanoid robot into a box?

[ LimX Dynamics ]

And now, this.

At the Opening Ceremony of WorldSkills Shanghai 2026 on 22 September, 19 Unitree humanoid robots performed alongside 120 dancers, presenting the world’s largest-scale performance featuring full-size general-purpose humanoid robots before an audience of more than 10,000 people, with a fully AI-driven autonomous robot cluster performance live-streamed worldwide in real time.

[ Unitree ]

A microwave box creates a surprisingly tricky vision system challenge: there’s a picture of a microwave on the side of the box, and that microwave has a window. From the camera’s perspective, it’s a box inside a box inside a box.

[ Pickle Robot Company ]

What exactly is Physical AI? Personally I think it’s just what robotics has always been, more or less, except that everything has to have “AI” in it now, so here we are.

[ MIT CSAIL ]

As the evolution of the UR Series, the g‑Series is designed to accelerate how robots are integrated into complete automation systems. It connects more directly with cameras, higher power EOATs, and control systems, eliminating external cabling and hardware to streamline integration. This enables cleaner and faster deployment of advanced automation.

[ Universal Robots ]

In Guadalajara, Mexico, many high schools have motivated teachers and talented students with an interest in science, technology, engineering, and mathematics, but they lack access to advanced tools such as robotics laboratories. The resources shortfall limits the students’ opportunities for hands-on learning on cutting-edge applications.

A team from ITESO, Universidad Jesuita de Guadalajara, is working to change that. Through the EPICS in IEEE initiative, a multidisciplinary group of 15 engineering students, faculty advisors, and IEEE Guadalajara Section volunteers developed RoboMeshA. The portable, self-contained educational platform brings robotics and AI experiences into classrooms.

EPICS is administered by IEEE Educational Activities and funded by the IEEE Robotics and Automation Society.

A mobile laboratory

Rather than requiring a school to build a dedicated computer lab or install complex software, RoboMeshA operates as an all-in-one mobile learning network.

“RoboMeshA brings robotics and AI to students who don’t have access to specialized facilities or preinstalled software,” says team member Fernando Vidal Luna, an IEEE student member and a mechatronics engineering major at ITESO.

Students connect directly to the platform from a user-friendly web browser. They can interact with the robot manually or use its control modes to watch it move and detect and avoid obstacles.

“The project combines mechanical design, embedded systems, control engineering, computer vision, and AI into a single robotic system that functions as a mobile learning laboratory,” says faculty advisor Jorge A. Lizarraga.

The team says young students are interested in technology, programming, and robotics but don’t have an opportunity to work with systems that combine mechanics, electronics, software, and control.

“RoboMeshA allows students to see how all these disciplines work together in a tangible and understandable way,” says team member José S. González, who also is studying mechatronics engineering.

The team has built two units and is developing a modular coupling framework to expand the system’s capabilities for research and classroom demonstrations. The structured system design approach connects independent software components while minimizing internal dependencies, enabling four RobotMeshA robots to operate together.

Overcoming design challenges

The team faced significant hurdles while designing the project.

“One key challenge involved the robot’s structural design,” Luna says. “It wasn’t only about making a chassis where all the components fit and the design had sufficient stability, rigidity, and weight distribution. It was also about ensuring that the electronics were protected while still being accessible for maintenance, testing, and modifications.”

“It was also challenging to design a platform that could be used by students with different levels of experience,” González adds.

“When students realize the technology they develop can inspire others and improve lives, engineering becomes far more meaningful.” —Luis Fernando Luque-Vega

The team partnered with the CETI Colomos and Prepa ITESO high schools to validate the platform in classroom settings.

“We wanted the first interactions with the robot to be simple and intuitive,” González says, “such that students could simply power the robot, connect to its network, and begin interacting with it, rather than having to deal with software installation, extensive configuration, or troubleshooting.”

Engineering with social impact

Many of the students who participated were from ITESO’s applied professional projects program. The experience offered them practical training in project management, system integration, and user-centered design.

The team also presented a research paper and a project poster in May at the Engineering Congress of the Jesuit University System.

“Seeing a design move from a digital model to a physical system was invaluable,” González says. “Working with students from different backgrounds taught us to listen to end users and design for their actual needs.”

Project lead Luis Fernando Luque-Vega, an IEEE member, says he’d like the venture to serve as a blueprint for engineering education.

“I hope RoboMeshA is adopted by schools, universities, and IEEE student branches across Mexico and internationally as a model for integrating technical innovation with community engagement,” Luque-Vega says.

By pairing engineering talent with community service, initiatives such as EPICS in IEEE demonstrate how targeted support can turn academic concepts into real-world solutions.

“When students realize the technology they develop can inspire others and improve lives, engineering becomes far more meaningful,” Luque-Vega says.

For more information on service-learning opportunities, visit the EPICS website.

Throughout her career, roboticist Barbara Mazzolai has turned to nature for inspiration. Now she wants to ensure the technology she builds gives back to the environment, too.

After starting her career as a biologist, a chance opportunity saw Mazzolai switch streams to engineering and become an early pioneer of bioinspired robotics. Building on her knowledge of biology’s ability to solve a diverse set of problems, she has developed robots based on octopuses, plant roots, and even seeds. “I’ve always been fascinated by living organisms, [and] by the extraordinary variety of solutions in nature, selected by the evolutionary process,” she says.

Barbara Mazzolai

Employer:

Italian Institute of Technology

Occupation:

Associate director for robotics; director of the Bioinspired Soft Robotics Laboratory

Education:

Master’s degree in biology, University of Pisa; master’s degree in eco-management and audit schemes, Scuola Superiore Sant’Anna; Ph.D. in microsystems engineering, University of Rome Tor Vergata

But Mazzolai, now the associate director for robotics at the Italian Institute of Technology, in Genoa, also believes engineering needs to reckon with its own impact on the natural world. That’s why she is advocating for a new field of research she calls “sustainability robotics.”

In a manifesto published in Nature Machine Intelligence in July, she and her collaborators outline a vision for a new approach to designing robots that’s meant to improve the relationship between nature, humanity, and technology.

“We need to reduce the footprint of our technology,” she says. “It’s really about thinking in a different way to open new possibilities for robotics [and] for society.” In this new mode of thinking, Mazzolai considers sustainability a core component of the design.

A child of nature

Mazzolai traces her fascination with the living world back to her childhood growing up on Italy’s Tuscan coast, close to the port city Livorno. Her father was a public-health inspector and a professional mycologist, and the family spent a lot of time exploring forests and learning about the local fungi and plants.

After toying with the prospect of pursuing art, her other major passion, Mazzolai ultimately decided to enroll at the University of Pisa in 1987 to study biology. She was particularly drawn to marine biology, but shortly before graduating with a master’s degree in 1995, she secured a research position at the Italian National Research Council’s Institute of Biophysics studying the cycles of heavy metals like mercury through both living and nonliving parts of the environment.

This involved collecting and analyzing samples from water, soil, vegetables, and even humans to understand the impact these metals have on health and the environment. She balanced this work with studying environmental management at the Scuola Superiore Sant’Anna, in Pisa, graduating with a master’s degree in 1998.

During that time, however, she learned that the university was recruiting biologists to help design new devices for environmental monitoring. She applied for and got the job in 1999 and began working as a research assistant under renowned bioroboticist Paolo Dario, first developing sensors and then robots meant to monitor air, water, and soil.

Even before entering a doctoral program, Mazzolai was promoted to assistant professor in 2004 and shortly afterward made her first foray into bioinspired robotics. In collaboration with colleagues at Sant’Anna, she helped design a soft robot inspired by the octopus. “We proposed it as a paradigm for launching this idea of soft robotics: demonstrating that [robots] can be soft, but at the same time apply strong force to the environment, like the animal does,” she says.

Back to school

In 2007 Mazzolai enrolled in a Ph.D. in microsystems engineering at Tor Vergata University of Rome, which she balanced with her role at Sant’Anna. She was already relying heavily on microfabrication techniques to develop sensors for her robots, and she was keen to push that part of the field forward.

While robots frequently feature sensors designed for perception, such as tactile or proprioceptive sensors, these systems typically focus on understanding the robot’s position in its environment, she says. “But there are few robots that integrate physical or chemical sensors to really understand the environment they move in,” she adds.

“I’ve always been fascinated by living organisms, [and] by the extraordinary variety of solutions in nature.”

Mazzolai was appointed as a team leader at the Center for Micro-BioRobotics of the Italian Institute of Technology in 2009, where she continued her work on the emerging field of bioinspired robotics. Two years later, she completed her Ph.D. and was promoted to director of the center.

Planting the seeds

Around this time Mazzolai says she became interested in using plants as a model for new kinds of robots, expanding bioinspiration beyond just animals. In particular, she was captivated by the ability of roots to efficiently explore the underground environment, and she imagined machines with the same deftness could have applications in both environmental modeling and precision agriculture.

While many bioinspired robots mimic animals, plants also serve as a muse for Mazzolai. This tendril-like bot can coil around other structures like a vine. Italian Institute of Technology

When she first proposed the idea, colleagues were somewhat skeptical of robots based on seemingly static organisms. But in reality, she says, plants move nonstop through a process known as indeterminate growth. “They really grow for their entire life,” she says. “They adapt their morphology, their behavior to the external environment; they repair, they sense, they communicate.”

Trying to mimic a system that operates on such different principles to conventional robotics required some serious thinking, however. Mazzolai says that working in bioinspired robotics sometimes requires you to have “two separate brains”—one of a biologist and one of an engineer.

The process often involves deep study of the target organism to learn the underlying principles that shape how it operates before trying to engineer a robot capable of mimicking them. “It’s not a copy of natural organisms,” says Mazzolai, because a living organism is both difficult to replicate and has different goals.

In the case of plant roots, what makes them so efficient at exploring the soil is that they reduce friction by growing only at the very fine tip of the structure, while the thicker base of the root remains static. This significantly reduces the amount of energy required to push through the earth compared to that of a more conventional drill, which must push the entire structure from above.

To realize this principle in a robot, her team developed a miniaturized 3D printer that sits at the machine’s tip and feeds thermoplastic filament through a heated nozzle to build a snakelike body behind it. This allows the robot to push through the soil efficiently. The tip also contains sensors that allow it to avoid obstacles and detect nearby nutrients or water.

Making robotics sustainable

After spending so much of her career borrowing from nature, Mazzolai is now eager to return the favor. Many modern technologies, including plastics and car batteries, have been developed with little thought about how they will affect the environment at the end of their life cycles, she says.

She wants to ensure that robotics doesn’t follow the same path. This is the inspiration for what she and collaborators now call sustainability robotics. The approach has three central pillars: ensuring that robots have minimal impact on the environment; that they’re available to people from across the world and all socioeconomic backgrounds; and that they’re “symbiotic,” providing benefits to both humans and nature.

More concretely, Mazzolai would like to incorporate the concept of a life cycle into the design of robots, so that at the end of their useful life these machines can be reused, recycled, or even biodegraded.

While that might sound ambitious, she’s confident that all the ingredients to make it a reality are in place. And it’s a vision that she is certain will inspire future roboticists. “There are younger people who want to really work in this field because this is the future, their future,” she says. Facing the threat of ongoing environmental damage, “they want to develop something that can help.”

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN

Enjoy today’s videos!

A quadrotor is really just two birotors that love each other very much.

[ CLIMB Lab ]

The last 30 seconds of this video are fantastic.

[ ETH Zurich Robotic Systems Lab ]

This is great! But if you really want to impress me, convince me that it’s a) safe to be in my home and 2) a realistic alternative to a human.

[ Figure ]

I doubt anyone would call Digit 5 pretty, but here is what serious humanoid safety looks like.

[ Agility ]

NYU Tandon researchers built an undulatory robot that pumps water between its head and tail to climb slopes, clear steps, and swim, all while keeping the same basic gait.

On an incline, shifting water toward the head increases the force pressing the robot’s front contact points against the surface, giving them more traction. Head-biased placement was the only configuration that successfully carried WorMa up the steepest incline tested, 19.5 degrees; every other configuration slid back down. It also reduced the robot’s cost of transport by at least 33 percent compared with the other configurations.
Steps required something different. Neither a fixed head-heavy nor tail-heavy robot could clear a step on its own. The researchers instead had WorMa approach the step with water in its head for traction, then shifted the water to the tail and raised the head and neck so the tail could push the robot closer. Once the head anchored on the step’s edge, the water shifted back to the head, and the robot continued. That sequence got WorMa over steps as high as 15 centimeters.

[ NYU Tandon School of Engineering ]

Kubi 2.0 is a desktop telepresence AI robot that allows remote users to look around and interact as if they were physically present. It is also a Physical AI device enabling AI agents to interact with people in the physical environment. For over the past 10 years in Japan, we have successfully implemented our telepresence solutions to help hospitalized and homebound students attend classes, interact with friends, and participate in school events. Now, we are expanding this initiative globally through Kickstarter.

[ Kubi 2.0 Telepresence Kickstarter ]

I am oddly impressed by this use case for a wheeled quadruped.

[ DEEP Robotics ]

Legged robots are versatile on land, but their use in underwater environments remains limited. Extending quadruped locomotion to water enables amphibious mobility with applications in inspection, environmental monitoring, and disaster response. This paper presents the design, modeling, and experimental validation of a reproducible underwater quadruped robot. The robot is built around custom waterproof motor housings machined from polyoxymethylene plastic, which use off-the-shelf O-rings and dynamic shaft seals. A simplified model is derived to describe the dynamics of this underwater legged system, capturing how drag forces on spherical end effectors transmit torque to the floating base. Building on this model, a closed-loop attitude controller is developed using an error formulation defined on the special orthogonal group SO(3). The controller is evaluated both in simulation and experimentally in a water tank, where the robot tracks desired orientation setpoints in roll, pitch, and yaw.

[ Norwegian University of Science & Technology Autonomous Robots Lab ]

Thanks, Kostas!

Ugo Nova, a domestically produced [in Japan] semi-humanoid robot, accelerates the social implementation of physical AI, from research and development to operation in industrial settings.

[ Ugo ]

I appreciate how much manipulation you can actually do with a good two-finger gripper and a clever robot.

[ Daimon Robotics ]

One humanoid robot rolls off the line every 10 minutes. UBTECH’s benchmark humanoid robot smart factory just goes live—from 1,000 to 10,000 units, we made the leap in less than 9 months.

This is great! But what are they all going to do?

[ UBTECH ]

MIT engineers have built a robotic optics lab that builds, runs, and dismantles an optics experiment on demand. The setup centers on a robotic arm that physically arranges mirrors, lenses, and other optical components that are each set in custom-designed, QR-encoded housings. The robot arranges each component in precise alignments to direct and focus a beam of light (red laser). The autonomous system could speed up testing of high-tech materials, from solar cells and sensors to video displays and quantum technologies. This video is sped up 2x.

[ MIT ]

Student-built rovers. Lunar terrain. High-stakes competition at NASA’s Kennedy Space Center. See the energy, engineering, and excitement of Lunabotics as university teams put their robotic systems to the test in a simulated lunar environment. These are highlights from the previous competition, as well as a look ahead as teams get ready for Lunabotics 2027.

[ Lunabotics ]

This CMU Robotics Institute Seminar is by Erol Şahin from METU, on “Augmenting Bee Colonies with Robotics and AI Technologies for Ecosystem Support.”

Earth’s ecosystems are facing a rapid decline in biodiversity, with honeybees—keystone pollinators critical to ecosystem stability—being among the most affected. The EU-funded RoboRoyale project addresses this crisis by integrating advanced robotics and AI to augment the beehive, enabling observation at unprecedented resolutions and scales. Featured on the cover of Science Robotics and receiving the 6th Edge of Government Award at the World Government Summit in 2024, our system tracks the Queen’s behaviors, colony efficiency, comb states, and long-term foraging activities, while advancing microrobotic intervention capabilities to support hive health. In this talk, I will discuss the challenges of developing this system, share key findings regarding complex social interactions, and explore the future potential of bio-hybrid research.

[ Carnegie Mellon University Robotics Institute ]

This article is brought to you by VicOne.

Robot safety has traditionally asked: Can a machine remain safe when something goes wrong? Physical AI raises a harder question: Can a machine remain safe when an attacker changes what it sees, decides, or does even when nothing appears to have failed?

As AI and robotics continue to advance at an unprecedented pace, modern robots perceive through multimodal sensors, interpret context using AI models, and translate those interpretations into physical action. As they move into dynamic environments, their safety increasingly depends on the integrity of the data guiding their decisions.

That dependence creates risks that conventional safety assessments may not fully capture. Recent research has demonstrated that manipulating what a robot sees, hears, or interprets can influence its behavior without requiring direct control.

Such manipulation can occur anywhere across its complex sensing and decision-making system — a layered attack surface encompassing training pipelines, system infrastructure, and runtime perception.

Layer One: Corrupting intelligence at its source

In 2017, BadNets demonstrated that a model could behave normally under most conditions, yet fail in the presence of a specific hidden trigger. In one example, a subtle pattern caused a stop sign to be misclassified as a speed limit sign without affecting the model’s behavior on other inputs.

What began as a classification vulnerability has since evolved into action manipulation.

At NeurIPS 2025, researchers introduced BadVLA a backdoor attack targeting Vision-Language-Action (VLA) models that allow robots to see, interpret instructions, and produce coordinated physical movement. Rather than altering a single label, the attack caused conditional deviations in the robot’s action trajectory when a trigger was present. Without the trigger, the model largely preserved normal task performance, while the backdoor remained effective under task transfers and model fine-tuning.

A related study in 2025, GoBA, showed that ordinary objects such as a coffee mug could serve as a reliable trigger. The researchers reported a 97 percent attack success rate without degrading performance on clean inputs.

A critical safety question today is whether Physical AI models remain within their task and safety boundaries under adversarial conditions.

These studies expose a blind spot in model validation: A model may pass testing yet produce corrupted behavior when a hidden trigger appears in operation.

So a critical safety question today is whether Physical AI models remain within their task and safety boundaries under adversarial conditions. Simulation tools such as NVIDIA Isaac Sim, when paired with VicOne Radeis, can test the effects of manipulated inputs before deployment.

VicOne LAB R7 demonstrates Radeis, a Physical AI safety validator for NVIDIA Isaac Sim that tests how adversarial visual inputs affect robot behavior before deployment.VicOne

Layer Two: System vulnerabilities as gateways to AI control

Even a securely trained model can be subverted if the surrounding system stack is vulnerable.

In September 2025, researchers disclosed UniPwn, a Bluetooth exploit chain affecting quadruped and humanoid robots from a major manufacturer. Hardcoded cryptographic keys allowed traffic decryption, authentication checks were bypassed, and command injection enabled root-level execution. The exploit is also described as “wormable.” A compromised robot could scan nearby units and potentially affect an entire fleet.

VicOne Lab R7’s demo shows how chaining three wireless exploits can trigger uncontrolled robot behavior within 60 seconds, resulting in operational disruption.VicOne

Middleware creates another exposure point. Vulnerabilities in ROS 2 and DDS-based systems can enable arbitrary code execution or abuse unauthenticated topics to deliver malicious commands. With sufficient access, an attacker could override motor commands or replace AI model weights without directly attacking the model architecture.

In this case, the components may still function as designed. What has changed is the trustworthiness of the commands flowing through the system. Vulnerability management can help teams identify known risks before deployment, while continuous monitoring can surface emerging threats.

Layer Three: Manipulating perception and reasoning at runtime

At runtime, manipulating inputs that shape perception or reasoning may require neither firmware modification nor a network breach.

In 2024, RoboPAIR demonstrated how carefully structured prompts could redirect LLM-controlled robots into unsafe trajectories. BadRobot exposed a deeper architectural weakness: in several cases, a robot verbally refused a dangerous command while its motion controller executed the action anyway.

Vision-based manipulation is equally powerful. VLAttack showed that an adversarial patch within the camera’s view could reduce a VLA model’s task success rate to zero. FreezeVLA showed that a single adversarial image could freeze a robot’s decision-making loop, making it unresponsive to subsequent instructions.

Runtime assurance must therefore look beyond whether individual components remain available and assess whether cyber events are beginning to affect physical behavior.

In each case, the camera may still work, the model may still run, and the controller may still respond. Yet the resulting behavior can be unsafe because the robot is acting on manipulated perception or reasoning.

Runtime assurance must therefore look beyond whether individual components remain available and assess whether cyber events are beginning to affect physical behavior. Security event correlation, behavioral-impact assessment, and policy-bounded response supported by edge AI, can help contain the affected path without unnecessarily stopping the entire robot fleet.

From point-in-time safety to lifecycle assurance

The risks across these three layers reveal the missing layer in robot safety assurance: cybersecurity. Functional safety addresses failures and unexpected operating conditions; cybersecurity extends that assurance to deliberate manipulation, including attacks that may leave the underlying system apparently functional.

This requires assurance across the robot’s lifecycle. During design, teams need to understand which cyber risks could invalidate assumptions behind intended behavior. Before deployment, they should test whether realistic attacks can cause a robot to deviate from its task or safety boundaries. In operation, monitoring should identify whether cyber events are beginning to affect behavior, contain the affected path, and preserve safe operation where possible.

VicOne’s lifecycle approach combines AI model and vulnerability scanning, simulation-based validation, and continuous monitoring to help secure robots from development through operation.VicOne

While cybersecurity does not replace functional safety, it helps ensure that Physical AI remains within acceptable boundaries even when what it sees, decides, or does is under attack.

For a deeper look at the cybersecurity risks and defense strategies shaping autonomous robotics, download our whitepaper “Securing the Rise of AI Robots: Cyber Risks, Real-World Threats, and Defense Strategies.”

Despite the recent deluge of videos of humanoid robots doing backflips and kung fu, the actual pace of progress towards humanoids that can do economically viable jobs at scale has at times seemed a bit plodding. There are some good reasons for this: A humanoid robot needs to be powerful enough to do useful work, safe enough for humans to walk past, and have enough battery life to limit how much of the day it’s hooked to a charger. But all three of those necessary features actively work against each other.

With its new Digit 5 robot, announced today, Agility Robotics may have found the sweet spot. Digit 5 is a humanoid worker that can lift 23 kilograms up 2.1 meters, can work in close proximity to people without relying on physical barriers, and can operate for more than 20 hours a day. This combination of power, safety, and battery life has resulted in a 1.8-meter tall, 129-kg robot that prioritizes functionality over style and pretty much everything else, but this is exactly what it’s going to take to get humanoids a job.

Humanoid robot companies have been reluctant to talk about safety, because there are no easy answers to “What happens if your fundamentally unstable bipedal robot falls over onto me/my pet/my baby?” And there are still no easy answers, but Agility does have an answer that allows Digit to be verifiably safe. That answer is to make sure that it’s physically impossible for Digit to fall over on anyone, ever, by “autonomously avoiding, stopping, or assuming a seated position,” according to today’s press release. In other words, as a person approaches Digit, Digit will put down whatever it’s carrying and then if necessary make sure that it’s stably seated on the ground before that person gets near. It may not be elegant, but it works, which is more than can be said for any other commercial humanoid that I’m aware of.

The other big change is Digit’s legs. Somewhat famously, Agility Robotics first introduced Cassie and then a whole series of Digit robots featuring bird-like ‘backwards’ legs. In his 2019 article for IEEE Spectrum, Agility co-founder Jonathan Hurst explained that this leg configuration was the result of a careful analysis of the physics of animal locomotion, rather than for the leg to look like any sort of animal in particular. The unfortunate reality for those of us who were fans of Cassie is that bird legs are optimal for dynamic motion, while human legs are better for squats and lifts, which will make sense to anyone who has seen an ostrich trying to lift a heavy box up off the floor and also anyone who has tried to outrun a cassowary.

How much will Digit 5 cost?

We also know a couple of things about Digit 5 that weren’t included in today’s announcement, thanks to Agility’s June SEC filing in advance of their plan to go public by the end of 2026. For example, at launch, Digit 5’s BOM (bill of materials) cost is likely to be somewhere between US $150,000 and $200,000. This is just the cost of the parts that make up a Digit robot, not what it costs for Agility to actually build one. Based on real production data, Agility is anticipating that with some near-term optimizing and at a volume of 10,000 units per year, that cost should drop to under US $50,000 per robot.

In this rendering, Digit 5 robots work in a factory with humans nearby, no safety barriers needed. When a human approaches, the robot puts down what it’s carrying and squats down so it can’t fall on anybody.Agility

Arguably less important than the per-robot cost is how Agility (and its customers) will profit from Digit 5. In the filing, Agility estimates (using “rounded estimates” which are “purely illustrative”) that Digit 5 robots will be offered to customers as a service at something like $8,500 per month. Based on 20 hours per day of work, and assuming that the total cost of a human worker to their employer is $30.50 per hour, Digit 5 as a service could save employers $100,000 per year, per robot.

It’s important to note that all of these numbers are just estimates, and that all kinds of things (many of them not under Agility’s control) could cause them to change. They’re most useful as an illustration of Agility’s broad approach to making Digit 5 profitable. It’s also important to note the assumption here with such a direct comparison is that Digit 5 will be a more or less effortless drop-in replacement for human labor, which is not something that I’m entirely sure has ever happened with any robot, anywhere.

Where does Digit 5 go from here? Today’s press release says that “as of May 2026, Agility had more than $300 million in multi-year customer orders for Digit 5, with a sales pipeline of prospective customers across manufacturing, warehousing, and logistics.” That works out to comfortably under 1,000 robots, one-tenth of the full capacity of Agility’s RoboFab factory in Oregon.

Between now and all those robots, however, lies an unpredictable entry into the stock market through what’s called a SPAC (special purpose acquisition company) merger. That’s expected to close within the next few months. Agility hopes to raise more than $620 million through the merger, and it will primarily spend the money to scale production of Digit 5 and get it to customers. Those in the European Union and United Kingdom should be able to buy the bots in 2027.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN

Enjoy today’s videos!

Traversing sparse 3D structures requires humanoid robots to perceive thin, overhanging geometry while executing agile, accurate whole-body motions. We study this problem through monkey-bar traversal, where the robot must jump to the structure, traverse it through sparse bar interactions, and land safely.

The list of obstacles that you can traverse to escape a robot is getting shorter.

[ ETH Zurich Robotic Systems Lab ]

YES GIVE ROBOTS TWO HEADS I LOVE IT!

[ General Robotics Lab ]

9/11 was the first documented use of robots for urban search and rescue and helped create the field of disaster robotics. Personnel began assembling on the afternoon of September 11 and worked the pile from late on September 11 through October 2, when the last available robot failed. The robots found no survivors, but they located remains and helped search for routes through the rubble toward basements and stairwells where trapped firefighters might have gone.

[ CRASAR ]

Unitree majorly fully open-sources the UnifoLM-WLA-1.0 embodied foundation model, achieving new SOTA results across multiple benchmarks among open-source models worldwide. A single model coordinates desktop and whole-body mobile manipulation, supporting cross-task and cross-end-effector generalization, driven by one model, whole-body coordination.

[ Unitree ]

Compliance is very important in physical interaction. In this work, we show how a multi-lined aerial robot uses its centroid and joint motion to achieve hybrid impedance—admittance control in contact-rich aerial manipulation tasks such as surface sliding. This work will be presented in IEEE IROS 2026.

[ DRAGON Lab ]

Thanks, Moju!

Remind me not to get too close to this.

[ RaiLab Kaist ]

Welcome to this edition of Things That Really Seem Like They Should Not Fly.

[ Texas A&M University Advanced Vertical Flight Lab ]

Achieving agile and generalized legged locomotion across terrains requires tight integration of perception and control, especially under occlusions and sparse footholds. Existing methods have demonstrated agility on parkour courses but often rely on end-to-end sensorimotor models with limited generalization and interpretability. By contrast, methods targeting generalized locomotion typically exhibit limited agility and struggle with visual occlusions. We introduce a unified reinforcement learning (RL) framework for agile and generalized locomotion that incorporates a novel attention-based map encoder in the control policy.

[ ETH Zurich Robotic Systems Lab ]

Finally, the killer app for humanoid robots! But we probably shouldn’t call it that.

[ Unitree ]

I suspect that this demo avoids many of the things that are actually difficult about doing dishes. Not just the water and the slippery soapiness, but also identifying when a dish is dirty as well as when it is actually clean.

[ Flexiv ]

Sure, I guess I might want a robot to deliver a burrito to me while I’m hiking to the top of a mountain in the rain...?

[ DEEP Robotics ]

AI has transformed the digital world. It writes our code, generates our images, reasons in our language. But the physical world—the plants that make our power, our fuel, our steel, and chemicals—it has barely touched. ANYbotics CEO and co-founder Péter Fankhauser on the bet behind the company: Why legged robots turned out to be the way into the world’s most demanding industrial plants, what it took to certify one for explosive atmospheres after experts called it impossible, and where autonomous industrial work goes next.

[ ANYbotics ]

Dexterous manipulation remains one of the biggest barriers keeping robots from successfully tackling a wide range of everyday tasks. A sense of touch could be the key, but a lack of quality data has held back progress. This is now starting to change as academic labs and startups race to build new tactile datasets and techniques to put them to use.

Over the last few years, vision-language-action (VLA) models have significantly improved the ability of robots to carry out complex tasks involving objects and environments they’ve never encountered before. Pretrained on huge amounts of images, video, and text, and then fine-tuned on a smaller number of teleoperated robot demonstrations, these VLAs can guide robots through a growing range of everyday jobs—like folding laundry, tidying living rooms, and even operating kitchen gadgets—using just a video feed and natural language instructions.

But robots still struggle with tasks that require fine-grained hand control, such as handling deformable materials or manipulating small objects—plugging in a USB cable or turning a key in a lock, for example. That’s partly because VLAs ignore one of the primary sources of information humans rely on in these situations: tactile feedback.

Manipulating Like Humans

“Most dexterous manipulation can be done by humans with their eyes closed,” says Trevor Darrell, professor of computer science at the University of California, Berkeley. “Understanding force, slip, and precise grasping is not something that can be done well with traditional vision sensors.”

However, making effective use of tactile sensors is difficult. Tactile sensor data has very different characteristics to the image data VLAs are normally trained on, and tactile datasets lag far behind the internet-scale of many vision and language datasets. To get around this, Darrell’s team devised a way to first pretrain a model on existing datasets before giving it a sense of touch by training a specialist submodel on 100 hours of specially collected, high-quality tactile data, including demonstrations of common actions like wiping, grasping, twisting, or pouring using more than 200 different household objects.

Explore an interactive visualizer of a small portion of the T-Rex dataset. T-Rex

Putting the tactile data to use was not straightforward. The goal was for a robot to be able to use the tactile signal to correct its grip in real time as it manipulated objects. But this requires reaction times faster than most vision-language models operate at. This mismatch is a significant challenge, Darrell says, so the team used separate submodels, known as “experts,” to handle high-level actions and low-level tactile control in a way that’s quick enough for the tactile feedback to be useful.

The action expert produces motion plans, while the tactile expert, which operates four times faster, uses tactile feedback to adjust the motion plan in real time based on what the robot is feeling as it goes. The model was then fine-tuned on about 100 teleoperated demonstrations of relatively complex manipulation tasks, such as screwing in a light bulb, applying toothpaste to a toothbrush, or transferring an egg between trays, where it averaged a success rate of 65 percent across 12 tasks—nearly double the best VLA model.

Data Diversity

One limitation, admits Darrell, is that his data comes from a single instance of robotic hardware. Robot hands range from fully articulated five-finger designs to simple pincer grippers, and tactile sensors can rely on fundamentally different physics, from measuring changes in resistance to recording images of a soft gel pad deforming. That makes most tactile AI research sensor-specific, says Chengbo Yuan, a master’s student at Tsinghua University in Beijing, and makes it hard to share data and transfer learnings between groups.

Yuan recently set out to tackle this problem by aggregating more than 3,000 hours of tactile robotic data from publicly available datasets, covering 21 sensor types and a variety of robot embodiments. Yuan says they were inspired by efforts like the Open X-Embodiment collaboration, which pooled data from many robots and led to models that generalize to hardware not used in training. Yuan’s team then designed a hardware-agnostic model that can train on this diverse data by converting each sensor’s output into a shared format and mapping it onto labeled positions on a template of a human hand. This model was much more successful than a baseline model, even on hardware it had never encountered before. Yuan puts that down to it acquiring “some kind of common sense of tactile knowledge,” by training on such diverse setups.

Chasing Scale

Despite the promising results, Yuan thinks more tactile data is needed, and his group is now leading an 80-institution collaboration to collate a larger set of teleoperated demonstrations using a standardized approach to tactile data collection and processing. In the meantime, Fudan University in Shanghai and its spin-out NeoteAI have already produced a tactile dataset an order of magnitude larger than previous efforts. Using a proprietary sensor attached to a variety of robotic arms and a handheld gripper operated by humans, they have collected more than 30,000 hours of demonstrations with synchronized visual and tactile data.

The researchers used this data to train a model that doesn’t just react to touch, but also proactively predicts what the robot should be feeling to help guide and assess actions, significantly improving performance. Shunlin Lu, a postdoc researcher at Fudan University and CTO of NeoteAI, says the results are clear evidence that access to large-scale and diverse tactile data leads to significant performance gains.

Robot manipulation policies with a tactile component offer improved performance on a variety of real-world tasks.NeoteAI

Another approach to scaling tactile data could be to piggyback on the vast quantities of visual robotics data already collected. Researchers at the University of Southern California, in Los Angeles, recently released a model that learned to infer tactile information from visual data, by training it on more than 2,700 demonstrations of everyday manipulation using a handheld gripper that records both tactile data and images from a camera on the device. The model learned associations between images of the gripper coming into contact with objects and the amount of pressure felt by the tactile sensors at that moment, giving even robots without tactile sensors a rudimentary sense of touch that the researchers showed to be particularly useful for contact-rich manipulation tasks. But their broader ambition is to use the generator to add tactile data to existing vision datasets.

How much tactile data will be required for breakthroughs in dexterous tasks remains unclear. So far, tactile training’s main contribution has been to make robots more efficient learners at tasks already within reach like picking and placing objects, says Yuan, and he suspects new algorithms may be required to tackle problems truly impossible without touch.

Long Cheng of the Chinese Academy of Sciences in Beijing also thinks raw data is no panacea. “Data is good,” he says. “But how to use them correctly is another issue.” The problem, he notes, is that vision provides a continuous, high-bandwidth stream of pixels, while tactile signals are sparse and intermittent, so models learn to ignore them. His solution, being presented at IROS 2026 later this month, is a model that predicts what a robot will feel from vision alone and then compares it against real tactile input. A large gap between the two means the sensor is detecting something the robot would otherwise miss, so these surprising signals are amplified while predictable ones are dampened. Across five contact-rich tasks, the approach averaged 62.8 percent success against 28.2 percent for the same model without touch.

Lu is more confident that data scaling could have similar benefits to those seen in areas like language and vision. He guesses closer to 100,000 hours, collected in varied, real-world settings rather than in the lab, could unlock new capabilities. Either way, the field now has some early signs that larger tactile datasets and smarter ways to use them can give robots a significant boost on some of the most challenging tasks. “I think tactile intelligence is actually the next step for physical AI,” Lu says.

You sit down and put your arm in the cradle. You press a button. The machine takes it from there.

A near-infrared light sweeps your inner elbow, hunting for a vein. A puff of alcohol hits your skin. An ultrasound probe glides across your arm, mapping how deep the vessel runs and which way it bends. Doppler captures the direction of blood flow to rule out the artery.

The cuff tightens around your upper arm. The needle comes down and pierces the skin. Your blood flows into the collection tubes, each one tipped end over end nine times—no more, no less. The needle withdraws. You get a bandage. No human ever touched you.

This is what it’s like to have blood taken by Aletta, the first autonomous blood-draw device authorized for use in the United States. Developed by the Dutch medical robotics firm Vitestro, the system combines imaging technologies with advanced robotics and AI to do by algorithm what a human phlebotomist—a trained health care professional who finds veins and draws blood by hand—does by feel.

The U.S. Food and Drug Administration gave Aletta the go-ahead on 19 August for use on adults in nonhospitalized settings. The decision follows the lead of European regulators, who authorized the device two years earlier.

“You have to tip your cap to them,” says Max Balter, a surgical-robotics specialist at Medtronic who worked on autonomous blood-draw systems during grad school in the mid-2010s. “The engineering that they have is incredible…and with their FDA clearance, it moves the whole industry forward.”

Aletta Boosts Lab Capacity Amid Shortages

In a clinical trial involving more than 1,600 people in the Netherlands, Aletta successfully drew blood on the first attempt in 94.5 percent of cases, even among those with hard-to-access veins, people with obesity, and the elderly. When Aletta failed to identify a suitable vein, the patient was referred for conventional phlebotomy.

“It’s exceptional performance,” says Joe El-Khoury, a clinical chemist at Yale who was not involved in the Dutch trial. “It’s definitely as good if not better” than a typical professional phlebotomist.

Complications were minimal, with multiple built-in safeguards to detect problems, such as sensors that track arm movement and needle position, and to halt the process should something go awry. And for those whose veins proved too challenging, a phlebotomist remains on hand to take over when needed.

Notably, because a single phlebotomist can supervise up to three Aletta machines, the system should go a long way toward “helping clinical labs address the critical operational challenges related to the staffing shortages of phlebotomists,” says Luuk Giesen, chief medical officer of Vitestro.

That’s no small challenge in a profession with a median annual turnover rate of nearly 25 percent and a vacancy rate of close to 10 percent, according to surveys of medical laboratories that draw mostly from U.S. institutions. The resulting staffing shortages can limit labs’ capacity to meet demand for routine diagnostic testing of blood counts, cholesterol, metabolic markers, and more. Aletta could address that bottleneck.

Addressing Skin Tone Bias in Blood Draw AI

The promise of greater capacity, however, comes with a caveat: Aletta still fails in roughly one case out of 20. Who are those people?

Some may simply have elusive or unusually deep veins. But a more significant obstacle could be skin pigmentation. In particular, the melanin in darker skin can interfere with the near-infrared light Aletta uses to first map the veins near the skin’s surface and identify promising puncture sites. The technique relies on hemoglobin absorbing the light differently from surrounding tissue, and darker skin tones can absorb more of that light before it reaches the camera, thereby reducing the contrast.

Some experts have raised concerns that Aletta’s infrared sensors will perform poorly for people with darker skin tones, but Vitestro says that its machine also includes an ultrasound sensor, in part to mitigate that risk.Vitestro

Giesen recognizes that the issue could affect first-pass imaging, but notes that the main determinant of vein selection and needle placement is the ultrasound system, which relies on sound rather than light and should not be affected by skin pigmentation in the same way. “Ultrasound is skin-tone agnostic,” he says, adding that Vitestro has unpublished data showing no effect of skin tone on the system’s performance.

The company thus claims on its website that the “technology works well for all skin tones,” an assertion echoed in the FDA press release announcing the authorization of Vitestro’s device.

But given the history of racial disparities in medical devices—particularly optical technologies such as pulse oximeters, which can be less accurate in people with darker skin and went largely unrecognized as a problem for decades—such claims warrant evidence, says El-Khoury, who has written about the issue.

Brooke Katzman, a clinical chemist at the Mayo Clinic who is collaborating with Vitestro, also wants more evidence that samples collected by the robot are as suitable for testing as those drawn by hand.

The Dutch trial reported little damage to red blood cells, but other measures of sample quality, including clotted tubes, insufficient blood, and proper tube filling, still need to be assessed, as do the results of routine laboratory tests themselves. Katzman plans to launch a U.S.-based trial next year to collect just that sort of data.

“We’re going to do our due diligence,” she says. “Like any instrument we would bring into the lab, we’re going to put it through its paces before using it clinically.”

The Future of Automated Blood Testing

Aletta takes its name from the 19th-century physician Aletta Jacobs, the first female doctor in the Netherlands and founder of what is widely considered the world’s first birth-control clinic.

That nod to history is fitting for a technology that builds on decades of research in robotic phlebotomy by groups in Europe, the United States, and China, and MagicNurse. Yet few pushed the concept as far as biomedical engineer Martin Yarmush of Rutgers University in New Jersey, in whose lab Medtronic’s Balter completed his Ph.D.

In one version of their platform, the Rutgers team even coupled their robot to a benchtop blood analyzer, allowing it to draw samples and then measure levels of infection-fighting immune cells and oxygen-carrying red blood cells—all within minutes.

That all-in system never made it out of laboratory testing. And VascuLogic, the company spun out to commercialize the platform, is long defunct—though others, including ROPHAI, BHealthCare, and MagicNurse, continue to work in the space. But the Rutgers proof-of-concept demonstration points toward the tantalizing possibility of fully automated blood testing at the point of care, with robots handling everything from the needle stick to the analysis.

It is, in some ways, the promise that Theranos made—but built on conventional, validated laboratory technology rather than the dubious science and deception that brought that particular company down.

“I have no doubt that is the future,” says Gregory Retzinger, a clinical pathologist at the Northwestern University Feinberg School of Medicine, in Chicago, who collaborates with Vitestro and has tried the Aletta device himself. (“It was painless, it was fast,” he says.)

For now, Giesen says Vitestro is keeping its ambitions—and its machine—focused on the blood-collection process itself, though he believes Aletta could ultimately do far more. The company plans to launch Aletta in Europe next year, with the U.S. market to follow.

This article appears in the November 2026 print issue.

Imagine you are trapped under rubble after an earthquake and you see an electronics-covered cockroach with a spring-loaded needle on its back scuttling toward you. Although the sight might be unnerving, to say the least, this prototype paramedic cyborg, or “Paraborg,” might one day help deliver lifesaving aid to disaster victims who might be otherwise impossible to reach.

The Hardest Problems in Robotics

For decades, scientists have sought to develop cyborg insects as “a shortcut around some of the hardest problems in robotics,” says T. Thang Vo-Doan, director of the University of Queensland’s Biorobotics Lab in Brisbane, Australia, which just published a paper on the Paraborgs.

The University of Queensland

Building an insect-size robot “that can move reliably through rubble, climb over irregular surfaces, recover from falls, carry its own power, and still have room for useful sensors is extraordinarily difficult,” Vo-Doan says. An insect already comes with much of that mobility built in, so instead of trying to create artificial versions of every part of an insect’s body from scratch, researchers can graft an electronic interface onto an insect to make use of its existing capabilities.

Previously, scientists have shown they could steer cyborg insects such as beetles and moths. This prior work largely focused on controlling their movements to serve as passive sensor platforms.

In 2023, as Vo-Doan and fellow researcher Thanh Nho Do were talking about search-and-rescue cyborg insects shortly before that year’s IEEE International Conference on Robotics and Automation (ICRA), they asked, “What happens after an insect finds a trapped victim?” Vo-Doan recalls. “Could it go beyond locating someone and actually provide some form of assistance while rescuers are still trying to reach them?”

Giant Cockroaches to the Rescue

To answer this question, the roboticists experimented with giant burrowing cockroaches (Macropanesthia rhinoceros), which are native to Australia. The researchers needed an insect capable of carrying a large payload (over half its weight), and at roughly 40 grams in size, this species is the world’s heaviest species of roach. A larger insect is also easier to operate on to implant cybernetic interfaces.

The cockroaches were saddled with lightweight electronics that included electrodes implanted into both their antennae and small tail-like appendages known as cerci. Wirelessly activating these electrodes with a handheld gaming controller could steer the roaches left or right, spur them forward, or stop them from moving.

The insects were also equipped with either a wireless camera or a remote-controlled injector, which used a spring to launch a drug-filled syringe at a nearby target. A chemical reaction inside the syringe then generated a puff of carbon dioxide, which exerted pressure within the syringe to inject its payload into a target.

Paraborgs are designed to work in teams, with some carrying cameras and others carrying injectors with potentially lifesaving medications.The University of Queensland

The scientists decided not to load both a camera and an injector onto a single roach because the combined weight and bulk could impair their mobility on complex terrain. Having both sets of electronics would also increase energy demands, resulting in reduced operation time. Instead, the researchers envision a swarm approach with the Paraborgs, with different specialized cyborgs performing complementary roles.

In proof-of-concept tests, the scientists were able to successfully navigate the Paraborgs over a 2.5-meter course past three checkpoints before launching their needles at an 8-by-10-centimeter silicone target. In 25 trials, the cockroaches completed the course every single time and succeeded at injecting the target 72 percent of the time. “The long-term goal is to combine the insect’s advanced locomotion with sensing and intervention capabilities so we can reach and help more people, more quickly,” Vo-Doan says.

While the Paraborgs can be steered remotely, the cockroaches themselves are still very much alive and able to use their skills as bugs to navigate through complex terrain.The University of Queensland

The researchers acknowledge that “for someone who is already trapped or injured, seeing a cyborg insect approaching could understandably be a little surprising or unsettling at first,” Vo-Doan says. Ways to make it clear these insects were part of rescue efforts might include flashing lights, recognizable emergency markings, “or perhaps a tiny speaker delivering a simple message such as, ‘Help is on the way,’ ” Vo-Doan adds. “Making people feel comfortable with the technology is just as important as making it work.”

Practical Paraborgs

In the future, Vo-Doan and his colleagues aim to address practical issues with the Paraborg. These include compensating for the movements of victims, establishing reliable wireless communications inside collapsed structures, and guiding the cyborgs as they climb over and squeeze through complex environments filled with rubble and dust. Autonomy will become increasingly important for these insects, particularly if the scientists want to operate multiple cyborgs at the same time, he says.

In addition, a fundamental challenge when it comes to working with cyborg insects is that they are living creatures with minds of their own. “We are not piloting them like conventional wheeled robots,” Vo-Doan says. “Electrical stimulation influences their direction, but the insect still generates and controls much of its own locomotion.” To deal with this unreliability, the Paraborgs will need better onboard systems to pinpoint their positions and monitor their actions so researchers can recognize when an insect has deviated from its course or become less responsive.

“We are not suggesting that this is a medical device ready to be used on people today,” Vo-Doan says. “Real disaster sites are full of unstable debris, narrow gaps, and communication difficulties, so we need to understand how the insect, electronics, and injector all perform under those conditions. There are also important questions around drug choice and dosage, sterility, needle safety, reliability, and regulation.”

Ultimately, such research into cyborg cockroaches may help inform robot design, explains Vo-Doan. These insects “can give us useful capabilities sooner while, at the same time, helping us develop the fully artificial systems of the future.”

The scientists detailed their findings last month in the journal Advanced Science.

This article appears in the November 2026 print issue as “Cyborg Roaches Could (Helpfully) Stab You.”

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN

Enjoy today’s videos!

I know these videos from Agility can be a little bit silly, but the couch drag in this one is impressive.

[ Agility Robotics ]

Stabilizing unsecured payloads against the inherent oscillations of dynamic bipedal locomotion remains a critical engineering bottleneck for humanoids in unstructured environments. To solve this, we introduce ReST-RL, a hierarchical reinforcement-learning architecture that explicitly decouples locomotion from payload stabilization. Successfully deployed on the Unitree G1 humanoid hardware, this modular approach demonstrates highly reliable zero-shot sim-to-real generalization across various objects and external force disturbances.

[ SteadyTray ]

Thanks, Ioana!

Figure is scaling compute so that its robots can...uh...have their compute scaled, I guess?

Solving for a robot in every home is not a data-and-compute problem, it’s a safety-and-cost problem.

[ Figure ]

The most important thing to know about this gripper is that koalas have two thumbs on each hand.

[ RAI Institute ]

DARPA Triage Challenge Finals are in November!

[ DARPA ]

Online, humanoid robots are very impressive to watch, but behind the scenes, most of those movements are carefully choreographed. Researchers in Carnegie Mellon University’s Safe AI Lab are instead teaching robots how to adapt. Their system, called APEX, allows a humanoid robot to navigate obstacles using adaptive, full-body maneuvers.

[ CMU ]

Researchers from North Carolina State University have created teardrop-shaped soft robots that leap upward or forward when exposed to infrared light—and will keep jumping as long as the light is present.

The robots are made of a liquid-crystal elastomer ribbon shaped like a teardrop, with a thin aluminum tube shaped like a V at one end. When exposed to light from an infrared lamp, the surface of the ribbon contracts, causing the ribbon to rotate. The stiff V at one end of the robot prevents the ribbon from simply rolling in place, causing the ribbon to twist tighter and tighter. This stores energy until the twist reaches a critical point when the ribbon releases that energy, causing the V at one end of the teardrop to snap downward and strike the surface. This launches the teardrop into the air.

[ NC State ]

Thanks, Ship!

I’ll be honest—I was prepared to be underwhelmed by the DARPA Lift Challenge, but there was such creativity in the heavy-lift drone designs that I’m excited for it to come back in 2028.

[ DARPA Lift Challenge ]

Thank you, Christian, for attempting to talk some sense into the internet.

[ Christian Hubicki ]

Humans use not only muscle signals but also stretched skin around joints as a cue for proprioception. To mimic this biological mechanism, we developed a three-layer joint-covering skin with 44 pressure- and stretch-sensitive elements for the musculoskeletal humanoid Musashi-W.

It’s not a replicant, but one day, it will be.

[ University of Tokyo ]

Thanks, Akihiro!

Having mobility issues with your robot? Just staple it to the end of an industrial robotic arm. Problem solved!

[ LimX Dynamics ]

But what if I am the sort of person who needs to speak to a manager?

[ Sharpa ]

In Turpan, China—known as the City of Fire—summer ground temperatures can exceed 50 °C. During the grape harvest, farmers traditionally carry heavy baskets back and forth under the intense heat, while every extra minute in the sun can affect the freshness of the fruit. This year, the DEEP Robotics Lynx M20S joined the harvest.

[ DEEP Robotics ]

This video showcases the achievements of the first OH! GYM! Project cohort, a group of university and graduate students who explored, developed, and deployed their own humanoid behaviors using the open-source AI Sapiens K1 platform. Over the course of one month, the students experienced the complete process of humanoid development—from creating motions in simulation to transferring them onto a physical robot through repeated Sim2Real experiments.

[ ROBOTIS ]

This article is brought to you by Tsubaki KabelSchlepp.

In modern automated manufacturing, six-axis articulated robots perform high-speed, multidirectional maneuvers under demanding operational cycles. However, as robot arms swivel, rotate, and extend, the electrical cables, fiber optics, and pneumatic hoses supplying them endure severe mechanical stress. Torsional twist, rapid acceleration, and repeated contact with machine structures often lead to premature conductor fatigue, insulation breakdown, and costly unplanned production halts.

To overcome these multi-axis motion challenges, the Tsubaki KabelSchlepp Robotrax System provides a specialized three-dimensional cable carrier engineered specifically for complex robotic motion.

Managing High Tensile Forces With Central Steel Technology

Conventional cable carriers often transfer operational movement stress directly onto internal electrical lines and hoses. The Robotrax system changes this dynamic through a central steel cable that runs through the core of every chain link.

The Robotrax system’s central steel cable absorbs the primary tensile loads and preserves conductor integrity, dramatically extending cable service life.

When robot arms undergo rapid directional shifts and accelerations up to 10 g, this internal steel cable absorbs the primary tensile loads. By isolating electrical and fluid lines from pulling forces, the design preserves conductor integrity and dramatically extends cable service life. Mechanics can easily calibrate and adjust system tension using an integrated clamping piece, ensuring consistent mechanical support throughout long operational cycles.

Spherical Link Design and Modular Cable Routing

The foundation of the Robotrax system lies in its open, single-piece plastic links featuring spherical snap-on connections on both sides. This geometry allows the carrier to flex smoothly across three axes, providing radial rotation of up to ±450 degrees per meter depending on the model size.

To optimize internal organization, carrier links contain up to three distinct chambers. This physical separation prevents signal interference and mechanical abrasion between heavy power lines, sensitive data channels, and fluid hoses. For standard models (R040 through R100), technicians can press cables directly into the carrier without tools, drastically reducing installation and maintenance time. Larger configurations, such as the R140X, incorporate swiveling crossbars with snap locks alongside vertical and horizontal dividers for customized interior partitioning.

​ROBOTRAX System

1. Steel cable for transferring extremely high tensile forces
2. Tension piece for locking the chain links
3. Type with toolless opening swivel crossbars and divider module available
4. Open design
   – Fast cable laying as the cables are simply pressed in
   – Easy checking of all cables
5. Special plastic for long service life
6. Protective covers or heat shields made from different materials are available for different environmental conditions
7. Quick-release bracket for fixing and continuation
8. Strain relief with LineFix clamps
9. Protection against hard impacts, excessive abrasion and premature wear as well as limitation of the bending radius through protector

Active Retraction and Impact Protection

Large robot work envelopes and high-speed motion trajectories can cause loose cable carrier loops to swing and strike the robot body. To eliminate these destructive collisions, Tsubaki KabelSchlepp integrates the Pull Back Unit (PBU).

The PBU serves as an active retraction mechanism that maintains optimal tension on the cable carrier throughout the entire motion cycle. By preventing excess slack and eliminating interfering contours, the PBU minimizes collision risks across complex movement paths. The unit requires zero maintenance on its retraction element and offers standard mounting configurations for leading industrial robot platforms, including KUKA, ABB, and FANUC.

Tsubaki KabelSchlepp’s Pull Back Unit maintains optimal tension on the cable carrier and minimizes collision risks across complex movement paths.

Additionally, external protectors can be retrofitted onto individual chain links. These durable impact shields limit the minimum bending radius to prevent over-flexing while shielding the chain body from severe external abrasion. If wear occurs, technicians simply replace the modular protector rather than the entire cable carrier assembly.

Built for Demanding Industrial Environments

From automotive welding cells to high-speed machining centers, Robotrax systems adapt to severe working conditions through tailored protective accessories:

• Heat Shields: Aluminum-coated textile fiber covers protect against radiated heat, hot weld spatter, and flying sparks.
• Protective Covers: Coated polyester sleeves shield sensitive lines against aggressive cutting fluids, hydraulic oils, paint overspray, and abrasive dust.
• LineFix Strain Relief: Multi-layer clamping devices anchor cables securely at both ends to prevent axial displacement during intense motion.

On a good day, the rock quarry in central Texas is about 370,000 kilometers (230,000 miles) from the moon. But last February, when Rishi Jangale watched his 1.8-meter-wide, 150-kilogram inflatable robot roll effortlessly over rocks, gravel, and wet clay, his imagination turned the quarry into the lunar surface instead.

Jangale is an upbeat mechanical engineer nearing the end of his Ph.D. at Texas A&M University, in College Station, Texas. He and his labmates have been working on this big tan “RoboBall” for about five years. Their goal: create a vehicle capable of exploring some of the most inaccessible terrain in our solar system, such as the 21 km-wide Shackleton crater on the moon’s south pole.

The Shackleton crater is 4 km deep and contains many smaller, deeper craters within. Some parts of the crater never see the sun, and within these perpetually dark, frigid pockets lie mysterious substances that planetary scientists have long struggled to examine, including layers of ancient lunar geology and stores of frozen water that could potentially support a future lunar base.

“The moon is like an archive of what happened to the Earth,” says Sara Russell, a cosmic mineralogist with London’s Natural History Museum who is not involved with Texas A&M’s work. “Robotic collection works brilliantly well, and it’s great to see that being explored more in this context.”

“NASA is not going to let astronauts get anywhere near these craters, because if someone falls in, you’re not going to be able to get them out,” Jangale says. “So we thought, what better shape to roll down a hill than a ball?” In a recent paper published in IEEE Transactions on Field Robotics, Jangale’s team reports on the design of RoboBall, a hypothetical lunar mission, and the results from initial tests in the Texas quarry.

Getting Rolling

RoboBall is the brainchild of Jangale’s advisor, former NASA robotics engineer Robert Ambrose, who first conceived of the design in 2003.

Ambrose figured that a ball could address a pesky mobility risk that robots face on lunar terrain, especially in low gravity: tipping over. An inflatable sphere can’t tip over, and the form factor also insulates its internal components from sharp rocks, dust, and the huge temperature swings from over 93 °C in sunlit spots to minus 240 °C in the shade inside lunar craters. Ambrose imagined a wheeled rover parking at a crater’s edge and releasing a RoboBall to explore its depths. Unlike a small tethered rover, the RoboBall wouldn’t roll its way back up—but with no strings attached, it would have a far greater range to collect geological samples, and then be able to launch them back to the rover outside the crater with small rockets.

A robotic rover would ferry RoboBall across the lunar surface to the edge of a crater.R. Jangale, D. Pravecek, et al.

NASA hasn’t brought lunar samples back to Earth in over 50 years. Some morsels of moon geology find their way to Earth as meteorites, but Russell says these lack the “gold standard” field work—context about where that sample actually came from. Even as NASA reboots its crewed moon missions, many lunar sites remain inaccessible.

In 2022, Ambrose’s lab finished a proof of concept, the 0.5-meter-wide RoboBall II. Creating the full-size RoboBall III then took about 11 months. “We were building these robots really quickly,” Jangale says. “Ambrose really encourages us to use and break these robots.” And designs did go awry. Jangale remembers software errors and a drivetrain that proved too weak to roll over soft bumps in initial tests.

Texas A&M RAD Lab/YouTube

How to Slow Your Roll

RoboBall drives by moving a pendulum within its shell, shifting its entire center of mass. On flat ground, if the pendulum’s arm points forward, the shell rolls forward to compensate, and as long as the pendulum keeps the center of mass in front of the center of the ball, RoboBall will keep rolling forward. If the pendulum leans a few degrees to the left or right, RoboBall steers left or right to match. “The robot wants to go where you point the pendulum,” Jangale says. The 340-pound ball is just soft enough to bounce lightly over bumpy obstacles, but its slight overpressure keeps it relatively firm. On steep slopes, this mechanism also lets the ball control its downhill speed by simply angling the pendulum uphill.

“The beauty here is the simplicity,” says Hiro Ono, an aerospace engineer who worked on robot mobility at NASA’s Jet Propulsion Lab for 13 years before joining Georgia Tech. For space robots, Ono describes simplicity in terms of the number of actuators. RoboBall has just two, and both are fully inside of the shell, shielded from environmental risk factors like dust, a feature that Ono describes as “unique.”

After the first quarry tests, it took about seven months for the team to design and build an upgraded RoboBall III with 2.5 times more torque—enough to fling itself over small obstacles and roll up 20 degree slopes.

A small rocket can launch out of the center of the RoboBall to return a sample back to a rover outside the crater.R. Jangale, D. Pravecek, et al.

Getting Mission Ready

In the new paper, the remotely operated RoboBall III descended the quarry’s slopes, navigated soft terrain, and launched hypothetical payloads back out of the crater with small rockets. Powered by a large battery, the robot would inflate itself during a lunar mission and charge up from a robotic rover at the edge of a crater before heading out on its own.

RoboBall is probably not the right platform for all kinds of missions—its slightly bumbling nature means that it’s not ideal if you want to collect a sample from a very specific rock, for example. For now, the team hopes to work with scientists designing lunar science instruments to physically fit into RoboBall’s carry-on-luggage-size interior, while also fitting in with how RoboBall operates. “The robot is not the mission,” Jangale says. “The robot is a way for you to complete the mission.”

The current version of RoboBall cost roughly $250,000 to build, but is not quite ready for the moon in its current form. While its gold-treated aluminum parts are appropriate for space missions, its other materials are not. Space-grade electronics will cost more, and the ball’s shell—made from a material tough enough to roll over steel shards and withstand minus 184 °C temperatures—has never been evaluated in lunar extremes. Aside from materials questions, the team plans to engineer the ball to adapt how it drives on varying slopes autonomously. They also hope to collaborate with government agencies or spaceflight contractors to keep refining the robot’s design for a real, eventual mission.

Later this year, Texas A&M will open a new facility in Houston with the world’s largest indoor simulated moon and Mars landscapes. The facility is only 190 km from Jangale’s quarry. It’s still about 370,000 km from the moon, but it’s going to help Jangale get his robot quite a bit closer.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN, TEXAS

Enjoy today’s videos!

Nvidia just paid US $12.9 billion for the company that acquired Pollen Robotics, and this must be why.

Meet Microduck. 🦆 The 25-centimeter, 780-gram robot that waddles, falls, gets back up, and learns new tricks.

Packed inside: 15 degrees of freedom, a front camera, an 8x8 lidar, two IMUs, mics, a speaker, NFC, Wi-Fi, and Bluetooth.

Out of the box, Microduck already walks, sits, crouches, roller skates, picks up objects with its articulated beak, and recovers from falls on its own. Drive it with a game controller, plug-in accessories, and NFC tagged objects, run autonomous behaviors, or gather several Microducks for races and football.

Software fully open source. Ready for whatever you throw at it.

On pre-order for an astonishingly low $399, and ships before Christmas.

[ Microduck ]

Thanks, Matthieu!

If you’ve chosen to ignore all the earlier DARPA Lift Challenge videos that we’ve posted, now you can get all caught up in about five minutes.

[ DARPA ]

You had me at “54-gram robot that out-jumps a kangaroo.”

[ IEEE Transactions on Robotics ]

Sometimes, you just need a video like this.

Most fish-inspired robots are built for one size and one job, so scaling them up or down usually means starting from scratch. A team of engineers says it has found a way to solve that problem. They’ve unveiled ScaFi, a robot modeled on fish like cod and mackerel.

[ New York University ]

Thanks, Leah!

Martin writes, “We’re a small robotics team in Czechia, Europe, building practical hardware around the Unitree G1. Here’s a short demo of our lightweight gripper picking up a strawberry; the gripper weighs under 200 grams and is designed for simple, sensitive manipulation without adding a complex multifinger hand.

[ Sentio Robotix ]

Thanks, Martin!

Hybrid visual markers that are useful for both cameras and lidar is a neat idea.

[ Hello Robot ]

Thanks, Binit!

EmoLo brings emotion-inspired expressive locomotion to Open Duck Mini V2, a low-cost, open-source bipedal robot inspired by Disney’s BDX droids. With a single reinforcement learning policy, the robot can generate distinct walking styles associated with different emotional expressions, showing how characterful and expressive whole-body motion can be achieved on an accessible robotic platform.

[ EmoLo ]

Thanks, Masato!

If it’s possible for a robot with a completely immobile face to look frustrated, this robot absolutely does, starting at three minutes into this video.

[ DLR RM ]

Noble Machines deployed its first general-purpose robots to a Fortune Global 500 industrial customer within 18 months of the company’s launch and met its first delivery milestone, made possible by its AI-driven whole-body control and industry-leading end-to-end autonomy.

[ Noble Machines ]

We’ve reduced the time it takes to go from physical prompt → robot behavior. The faster anyone can teach a robot to do something new, the easier it becomes to scale physical work.

[ Generalist ]

I know this video is mostly a gimmick, but I would totally rent a moderately heavy lift quadruped for a couple of days to help with a move.

[ DEEP Robotics ]

Is taking two minutes to excellently fold a shirt too long, or do we even care how long it takes, as long as it’s a robot doing it?

[ Tokyo Robotics ]

TRON 2 × Wuji Hand 2 handles TCM pharmacy work: picking, weighing, grinding, and packaging. The omnidirectional base frees the hands, while precise gripping and dual-arm force control enable midair operations.

[ LimX Dynamics ]

Person who genuinely knows things about robots, Christian Hubicki, explains everything about robots smashing into walls.

[ Christian Hubicki ]

This article is brought to you by Ollobot.

From about 2017, individuals began to truly connect with the initial wave of companion robots. These devices had personality, moved around, joked, and answered when you spoke to them. Most early companion robots, however, were still limited by simple voice-command interactions and narrow functionality. Once the novelty wore off, many ended up sitting unused on shelves. As some of those companies went out of business and turned off their servers, many owners likened it to losing a pet.

What Ollobot describes as “gentle intelligence” is a useful way to think about where the serious work in this category is going. Not toward more powerful assistants, but toward more present ones.

The problem companion robots were trying to solve

Loneliness is not a niche issue. According to one study, nearly one out of three elderly adults resides alone, meaning they do not have daily companions. Research also shows that children whose parents have migrated for work, leaving them in the care of relatives, were 2.5 times more likely to experience loneliness than children whose parents remain with them. Among working adults living alone in urban environments, similar patterns of social isolation emerge, even if they are less visible.

Over the years, technology has time and again attempted to solve this problem via video calls, smart speakers, and messaging apps without much success. Those tools are geared towards communication between people that already have relationships. They do not create presence. They schedule it. That is the gap that a new generation of AI companion robots is being engineered to fill.

Today’s AI robots are different

Today’s companion robots are not just cute and cuddly. They are designed with psychological research, clinical insight and long-term interaction models to be truly useful in real homes.

Three fundamental shifts define the current generation:

1. From reactive to proactive response. Older robots relied on you speaking to them, but modern robots monitor a room with cameras, microphones, and surroundings sensors to initiate interactions without your input, and they can pick up on your emotions.
2. From function-oriented to emotion-oriented design. The original pitch for companion robots was about what they could do. The question driving the serious work now is how they make you feel, which is a harder engineering problem and a more honest framing of what the product is actually for.
3. From standalone hardware to connected ecosystems. Leading brands are creating platforms rather than devices with software included as a built-in layer and remote access from the beginning.

The global AI companion market size was valued at US $36.8 billion in 2025 and is projected to grow from $48 billion in 2026 to $318 billion by 2033, at a compound annual growth rate of 31 percent from 2026 to 2033.

Three household scenarios and interaction models

Ollobot’s advanced AI family companion robot OlloNi SS1 addresses a number of gaps in what existing technology offers.

Elderly individuals living alone. The combination of proactive interaction, fall detection, and persistent presence addresses both safety and companionship without the social overhead of asking family members to check in more frequently.

Children in households where parents work far from home. The SS1 functions as a consistent companion that already knows a child, their preferences, their moods, and their routines. The remote connection features allow parents to stay present without requiring a scheduled call, and the life recording system gives them a passive window into their child’s days that feels less clinical than a monitoring camera.

Single professionals living alone in cities. The SS1 adapts to daily routines, builds up a preference model over time, and provides ambient social presence without demands.

OlloNi SS1 adapts to daily routines over time.Ollobot

What OlloNi SS1 is doing differently?

Ollobot’s goal in building intelligent companion robots is to address the gaps in technology and capability, using innovation not to automate tasks but to fill emotional voids.

Much of the robotics industry has historically pursued human imitation — machines that speak, look, or behave like people. The SS1 is instead designed around familiarity and long-term coexistence rather than realism.

The system integrates multiple subsystems operating in parallel, including visual perception, audio processing, mobility control, and interaction management. It is equipped with a multi-chip AI 4K vision module capable of facial recognition and motion tracking. One small but revealing detail is the inclusion of a physical privacy cover for the camera — a mechanical solution to concerns that software settings alone may not fully resolve.

OlloNi SS1 can actively integrate into family activities, and it can autonomously move closer to capture memorable moments or reposition itself to remain engaged in ongoing interactions.Ollobot

The robot supports advanced mobility across multiple indoor surfaces, including wooden floors, ceramic tiles, and low-pile carpets, with slope climbing capability up to 3.5 degrees. Rather than remaining in a fixed location, it can move naturally throughout the home to stay close to household members as daily activities unfold.

For example, the OlloNi SS1 may greet family members when they arrive home, follow an older adult from the living room to the kitchen while continuing a conversation, remind a child to take a study break after a prolonged period of inactivity, or notice that someone appears unusually quiet and gently check in. During family activities, it can autonomously move closer to capture memorable moments or reposition itself to remain engaged in ongoing interactions.

The robot continues to evolve over time, with over-the-air updates that deliver new features, performance improvements, and AI enhancements

It also incorporates fall detection with optimized accuracy for safety monitoring scenarios. A 6-microphone array enables omnidirectional voice pickup with an effective voice capture range of up to 5 meters, supporting reliable wake-word detection and far-field interaction.

To support continuous companionship, much of the robot’s AI processing takes place directly on the device through its “heart module” architecture, with 16 GB of memory and 64 GB of local storage. This enables the system to retain household memories, recognize familiar faces, and respond with lower latency, making interactions feel more natural even during everyday routines.

Because companion robots are expected to remain available throughout the day rather than only during brief interactions, the SS1 is designed for extended operation, offering up to 12 hours of standby time and around 5 hours of active interaction on a single charge. This allows it to accompany users through meals, conversations, playtime, and other daily activities without frequent interruptions.

To support engaging interactions, much of the robot’s AI processing takes place directly on the device through its “heart module” architecture.Ollobot

Like the relationships it is designed to build, the robot continues to evolve over time. Running on Android OS with over-the-air (OTA) updates, the system continuously receives new features, performance improvements, and AI enhancements, allowing its capabilities to grow alongside the household it serves.

The robot’s behavioral model also improves over time. Rather than reacting to isolated commands, it attempts to establish a baseline understanding of household routines and individuals. Changes in behavior — prolonged quietness, unusual inactivity, or emotional cues — become triggers for interaction.

Presence instead of utility

Several features in the OlloNi SS1 illustrate this emphasis on presence and continuity in its interactions.

The system can identify different household members, including pets, and adapt responses accordingly. Remote communication features allow family members to connect through the device without treating every interaction like a scheduled call. Environmental sensors support contextual reminders tied to weather or room conditions.

Its “2+1” multi-display configuration is also designed around emotional communication. Two circular side displays function as expressive “emotional eyes,” while a separate primary display handles information and structured interaction. The separation allows emotional signaling and functional communication to operate independently, creating more intuitive nonverbal interaction even when no dialogue is taking place.

The robot’s behavioral model improves over time. Rather than reacting to isolated commands, it attempts to establish a baseline understanding of household routines and individuals.Ollobot

The SS1 also includes an automated life-recording system built on facial recognition and behavioral-event detection that can capture moments such as laughter, physical closeness, or group interaction automatically. An integrated AI vlog engine can then organize those moments into edited short-form videos with automated sequencing and soundtrack generation. The design intent is to preserve spontaneous domestic moments without requiring active documentation behavior from users.

An integrated AI vlog engine can organize recorded moments into edited short-form videos with automated sequencing and soundtrack generation

Visual data is processed primarily on the device through the SS1’s on-device AI architecture, with household memories stored locally and managed within Ollobot’s proprietary ecosystem instead of being shared with third-party smart home platforms. Access to recordings and live feeds is restricted to authorized users through the companion app, while encrypted communication helps protect data during remote access. Users also retain direct control over recording preferences, and the physical camera privacy cover provides an additional hardware-level safeguard whenever visual monitoring is not desired.

Remote communication is similarly structured around persistence rather than transaction. Traditional video calls are episodic and screen-bound; the SS1 instead acts as a continuously present interface embedded inside the household environment. Through autonomous mobility, environmental awareness, and persistent household memory, remote family members interact with an ongoing domestic context.

The larger shift to “gentle intelligence”

Ultimately, gentle intelligence is not about making robots behave more like humans — it is about helping them fit more naturally into human lives. Each OlloNi SS1 unit develops a unique behavioral profile based on its household. Two units running in different homes for a year will have become meaningfully different from each other, shaped by the specific people, habits, and rhythms of where they live.

That kind of long-term personalization is what early companion robots never had. It is also what makes the difference between a product that ends up on a shelf and one that actually earns its place in a home.

Learn more at ollobot.com.

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

IROS 2026: 27 September–1 October 2026, PITTSBURGH

CoRL 2026: 9–12 November 2026, AUSTIN

Enjoy today’s videos!

This is very, very cool. But I’m trying to think of what the commercial use case will be, you know? I guess, high speed, incredibly dangerous package delivery to second-floor windows or something...?

[ Unitree ]

Humans have a remarkable ability to perform new physical skills from only one or a few examples. Our latest robot foundation model, GEN-1.5, exhibits the beginnings of that same ability: It can learn a new task in seconds, from a single example, without gradient updates or fine-tuning. It displays broad capabilities across one-shot and few-shots learning from demonstration, as well as zero-shot physical generalization. Although the tasks are simple and short-horizon, this is the first model we know for which one-shot and few-shots learning of physical skills have emerged at scale. We view these results as a significant step toward our mission of building general intelligence for the physical world.

I will make the cautionary point that for many of these “the model figured it out” tasks, the blog post can only say that there was no relevant pretraining data “to the best of our knowledge.”

[ Generalist ]

BeanBot is a robot inspired by Mexican jumping beans, and I need say no more.

[ IIT ]

As a professional bagpiper who definitely pays very close attention to whatever that annoying tapping noise is coming from the back of the band, I can attest to this group of robot drummers being absolutely top-notch.

[ AgileX Robotics ]

What does it take for an aerial robot to move through a sequence of arbitrary poses—fast, precisely, and continuously? Rather than teaching the robot a behavior from data, we asked how far a first-principles analytical model could take us. Through a collaboration between the AIMS Group at the Hong Kong Polytechnic University and DRAGON Lab at the University of Tokyo, we developed the first sequential-convex-programming-based trajectory-optimization framework for generalized multirotors, covering both conventional and omnidirectional platforms.

[ DRAGON Lab ]

Thanks, Moju!

This is a nifty idea that adapts a kind of interface frequently used for robot training and uses it for human training instead.

[ MIT ]

Gravis Robotics brings robotic intelligence to heavy construction machines. Our retrofit kit, the Gravis Rack, turns off-the-shelf hydraulic machines into robots. Cameras, lidar, and onboard compute lets your machine see and understand its surroundings, and learning-based control lets it work close to its limits, moving more dirt with full, fast cycles.

[ Gravis Robotics ]

Robust brachiation requires precise hand movements to grasp and release bars together with highly coordinated whole-body motion. To address this challenge, we propose a learning-based framework centered on waypoint-guided reinforcement learning (WGRL). WGRL guides the end effector through waypoints while allowing RL to explore and generate dynamic whole-body behaviors. With this approach, the learned policy demonstrated robust brachiation across diverse courses with different bar heights, spacings, and orientations in sim-to-sim experiments. In the real world, our life-size dual-arm robot successfully traversed four consecutive bars.

[ EVARL ]

Thanks, Ayumu!

Well, here’s a different approach to welding in shipyards with robots.

[ Kawasaki ]

We should have a lot more robots in agriculture, if only they’d lettuce.

[ Flexiv ]

We’ve all had refs like these.

[ PHYBOT ]

I got stuck after the first 15 seconds of this video trying to imagine what any of these home humanoids would usefully do if they dropped a glass.

[ Zhejiang Humanoid ]

Shakey the Robot doesn’t get enough love.

[ SRI ]

This work introduces a novel approach to physical human-robot interaction (pHRI) by leveraging the joint torque sensors of standard collaborative robots. By mounting a passive, uninstrumented plexiglass touchpad to the robot’s flange, we transform the robot into a handwriting-based input interface.

[ TS-Robotics ]

This article is part of our exclusive IEEE Journal Watch series in partnership with IEEE Xplore.

Microspines are one of many ways to enable robots to latch onto surfaces like walls and ceilings. Now roboticists in Canada are using the mini spikes to get drones to land on a more challenging, remote surface: icebergs.

Like a spider, the Ice Dart can land on and latch onto steep, slippery surfaces such as icebergs and glaciers—an increasingly useful capability as activity in the Arctic increases. The drone can grip onto icy slopes of nearly 60 degrees, which is way beyond what most humans could manage without special equipment.

In a recent study, researchers explained how they developed the Ice Dart drone with a special landing gear that absorbs the impact of a hard landing while holding the drone in place with tiny spines that penetrate and grip the ice.

Published in IEEE Transactions on Field Robotics, the study describes how the Ice Dart was able to land on icebergs and a glacier in southeast Iceland. Tests took place amid persistent winds and temperatures of 0 to 10 °C along the ruggedly breathtaking Fjallsjökull (pronounced “FYATLS-yuh-kuutl”) glacier, which empties into a lagoon filled with icebergs. The drone was able to successfully perch at speeds of up to 3 meters per second and slopes of up to 58 degrees, with a success rate of 100 percent even in wind speeds of 30 km/h.

The researchers were motivated by a desire to allow drones to land almost anywhere in the world, since the availability of safe landing sites is one of the primary limitations on where and how drones can operate. The researchers already have a history of developing drones that can land on fast-moving trucks as well as trailers, boats, and steep roofs.

Ice Perching

“The ability to land rather than hover can fundamentally change how drones are used in the field,” says Alexis Lussier Desbiens, a professor of engineering at Université de Sherbrooke, in Sherbrooke, Quebec, Canada, who coauthored the study. “Once a drone has landed, energy consumption drops dramatically, allowing much longer observation periods with a small aircraft. The drone also becomes completely silent and can even reduce or eliminate its thermal and RF signature by shutting down major onboard systems.”

Landing on icebergs specifically allows drones to monitor them for days or months, producing more detailed observation than a quick aerial surveillance mission. This could simplify iceberg tracking compared to methods such as helicopter deployment, dropped instruments, or dart-like tracking devices, and provide another data layer to satellite and ship-based iceberg detection, according to the researchers. It could also provide a means of monitoring icebergs that are otherwise untrackable.

With its carbon-fiber construction, the Ice Dart drone weighs just 2.65 kg and has four legs arranged in an X shape, attached to its body with a pivot joint. Used in the group’s previous drone research, this landing gear disperses energy to reduce impact and overcomes multiple engineering challenges. The friction shock absorbers consist of 38 disks that generate friction torque as the legs move up and down upon impact. This lowers the UAV’s center of mass and helps spread out the kinetic energy of landing, but the real trick comes in the form of two retractable spines on each foot—one for uphill and one for downhill grip. The larger spine engages on the more heavily loaded downhill feet, and the smaller, thinner spine engages more easily on the uphill feet, even under very low loads on steep slopes. The spines only penetrate the ice as the suspension compresses, generating grip and protecting them from high-impact forces.

“The inspiration for the retractable spines in the feet came from looking at a cat’s claws and their ability to deploy only when needed,” says Isaac Tunney, a Université de Sherbrooke postdoc in mechanical and robotics engineering who was lead author of the paper. “I wanted to create feet that would naturally and passively engage their spines in the ice at the right moment, regardless of the drone’s orientation, the surface geometry, or the ice conditions.”

Arctic Surveillance

William D. Harcourt is a researcher at the University of Aberdeen, in Aberdeen, Scotland, focused on Arctic glaciers, snow, and sea ice, as well as the use of remote sensing and machine learning techniques. Harcourt was not involved in the study, but he sees several potentially interesting applications of the technology.

“Near the front of tidewater glaciers, these systems could enable measurement of stress and strain and help us understand calving processes,” Harcourt says. “Drones can be used as a mobile GPS, literally acting as a receiver on the ice, but the system would need to solve tilting issues as 3D change measurements usually required the antenna to be horizontal. However, if these problems can be solved, it could be used to track iceberg movements.”

The researchers want to continue developing the Ice Dart technology for real-world applications, including autonomous landing site selection and an emergency takeoff capability to be used if an iceberg rolls over or breaks apart. This August, the drone will be deployed during a Canadian Arctic mission to land on icebergs, collect data, and help validate ship-based iceberg-detection systems.

Humanoids desperately need to stop making YouTube videos and get a job already, and Persona AI is one of the few humanoid companies that seems to be entirely focused on making that happen. Persona AI’s approach has been to carefully select a job that is economically viable for robots right now, and they’ve found one that was also the job of one of the very first industrial robots ever sold: welding.

IEEE Spectrum first spoke with Persona two years ago, shortly after it was founded by Nicolaus Radford and Jerry Pratt. Radford led the Valkyrie program at NASA’s Johnson Space Center back in the day and was also the founder of Nauticus Robotics, while Pratt led IHMC’s DARPA Robotics Challenge team before spending a couple of years as CTO of Figure.

The Challenge of Humanoids

As of our first conversation in 2024, Persona had committed to building an economically viable humanoid, but they hadn’t yet figured out where their focus was going to be. “We were all over the place,” Radford says. “Warehousing, automotive, we probably even mentioned the home.” These are the same environments with the same sorts of potential applications that basically every other humanoid robotics company is attempting to make economically viable, and despite an ever more exhaustive number of demonstrations, so far none have succeeded at any sort of useful scale.

The challenge for Persona, and really for every robotics company, is that it’s not enough that you have a robot that is simply capable of doing a task. It’s also not enough that your robot can do that task in a way that is efficient, reliable, and safe. What’s required is that your robot can make money for both you and your customer. Most humanoid companies seek to achieve this by targeting baseline “unskilled” human labor.

Persona did not see economic viability in the unskilled labor approach, Radford says. “We started forming this thesis around skilled trades and tool usage.” Persona is targeting much more expensive skilled labor with its robots, and the reason why this is feasible is because their entry point focuses on the kind of skills that robots are especially good at. “I like to call it ‘last-mover advantage.’ We’ve seen everything that everybody’s doing, and we’ve decided that there’s a different way.”

Persona AI/YouTube

A Humanoid for Shipyard Welding

The first task that Persona’s humanoid is focusing on is welding—using a handheld tool to connect one piece of metal to another. “Tool use is pretty difficult,” Pratt says. “And we want to use the same tools that humans do, which makes it more difficult.” That difficulty is offset somewhat by the fact that Persona’s humanoid will first focus on making long, linear welds that are relatively uncomplicated. “This is not the hardest style of weld,” Radford says, “but in shipbuilding you need a lot of them—hundreds of kilometers of linear welds per ship.”

Currently, Persona has two public partnerships: one with HD Hyundai, which is the world’s largest shipbuilder, and the other with POSCO, one of the largest steel producers in the world, both in Korea. Persona declined to get into detail, but Radford says that broadly speaking, the company is interested in customers who can support ‘hundreds’ of robots per location.

Shipyard welding is an enticing application for Persona because there is a deficit of skilled (and highly paid) workers, it’s taxing physical labor, and it’s a comparatively easy skill for a humanoid to learn.

The welding process is skilled in a very robot-friendly way. Because you can only weld as fast as metal melts, the top speed for the task is an easily manageable centimeter per second. And making a high quality weld involves millimeter-scale repeated motions, which robots excel at, especially over long periods of time—whereas humans tend to get tired or bored. Pratt expects that for these uncomplicated welds, performing on par with humans—if not eventually better—will be achievable soon.

Shipyards make a compelling case for a humanoid with legs, as opposed to a more stable wheeled base. “These open-air shipyards are a couple hundred meters long, with horizontal and vertical spars that you have to step over all the time,” Radford says. “You’ve got to work on the ground, overhead, and through portholes.” Persona considered other form factors, like four legs (or even more), but determined a two-legged robot would be the least disruptive to existing shipyard rhythms.

The Economic Viability of Humanoids

Deploying their robots in shipyards specifically brings additional advantages for Persona. The safety concerns that come with bipedal robots—such as potentially falling over on a human worker—are lessened because a shipyard environment is staffed with workers who are trained to work around potentially dangerous industrial equipment. The company is also less sensitive to competitive pricing because no other company is pursuing the use case. “We’re now in an industry where the value added by our robot can be high enough that we don’t have to cut corners on quality and features in order to reduce the price,” Pratt says. “With a robot for the home, for example, there would be a lot of competition and a ton of price pressure.”

The added value for shipbuilders, Radford explains, doesn’t come from replacing humans with robots. “Our current partnerships are running at a significant backlog, and they’re labor-constrained. So we want to help our customers’ top line, not necessarily their bottom line.” In other words, rather than trying to argue that their robots will lower shipbuilding costs, Persona is instead arguing that their robots will allow more ships to be built. “Even if our robot was more expensive than a human, that would still be valuable to these companies, because it could unlock additional revenue,” Radford says. And when the additional skilled labor does not exist, Persona’s robots could be the next-best option for shipbuilders who need to scale.

Shipyards are environments where legs are necessary for a robot to be useful.CFOTO/Future Publishing/Getty Images

Persona’s Multipurpose Future

In the current commercial humanoid climate, where the emphasis seems to be on developing a “general purpose” robot (whatever that means) that will somehow justify itself through some undefined scale in some equally undefined and perpetually receding future, Persona stands out with their focus on a seemingly viable, near-term, and very specific business case. It hasn’t been easy, though. “It hurts us a little bit,” Radford says. “We’ve been told that we’re not thinking big enough.”

But a tool-using heavy industrial humanoid has plenty of future applications, many of which can be expanded from the welding skill even within shipyards. “Shipbuilding is a great beachhead,” Pratt says. “There are tons of adjacent markets, like grinding, painting, and other kinds of fabrication.” Persona’s ambition, Radford adds, is to be “the largest repository of industrial skills.”

It’s going to take time to get there. That time will be needed to collect tens of thousands of hours of expert demonstration data, create high-fidelity simulations, and conduct real-world testing. And however promising Persona’s approach may seem, the company still has to prove that its idea for an economically viable robotics company can be realized. It’s the same challenge that every humanoid robot company is facing. “A lot of the technical problems are the same no matter whether you’re in a house or a shipyard,” Pratt says. “Everybody’s got a great team and smart people, and we’re all knocking these problems out together.”

Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.

Actuate 2026: 18–19 August 2026, SAN FRANCISCO

IROS 2026: 27 September–1 October 2026, PITTSBURGH

Humanoids Summit Seoul: 22–23 September 2026, SEOUL

Enjoy today’s videos!

Speaking from experience, I can tell you that the best part of any DARPA challenge is when things go horribly wrong. And after you enjoy all the crashes (followed by all of the battery fires), get caught up with the DARPA Lift Challenge with video recaps of the final few days.

[ DARPA Lift Challenge ]

Drone delivery: coming soon to a moving vehicle (or perhaps even through an open window) near you.

[ HKUST Aerial Robotics Group ]

This tiny little robot called STEMbot (as in stem, not STEM) can climb up and around plant stems to check for pests. It’s not very fast, but it sure is adorable.

[ STEMbot ]

Monumental’s robots delivered the brickwork for a semi-detached home, laying around 20,000 bricks in a new community.

[ Monumental ]

Meet the world’s most “truss’t-worthy” robot.

[ Modlab University of Pennsylvania ]

Stanford BDML and Honeybee Robotics propose a payload to test gecko-inspired adhesives in spaaace!

[ NASA ]

How can a legged robot organize its own walking while maintaining a desired direction? In this work, we present a Differential Adaptive Steering (DAST) mechanism for directional adaptation in legged robots under decentralized adaptive control.

[ BRAIN VISTEC ]

I do not care even a little bit if a robot fails (safely, of course), as long as it recovers from that failure.

[ Sanctuary AI ]

Even for a robot that doesn’t drink champagne, those are some pretty light pours.

[ Kawasaki Robotics ]

If we as a society would just accept that the appropriate place to store clothing is in a pile on the floor, robots would have a much easier time of it.

[ LimX Dynamics ]

To be fair, this is also the speed at which I fold shirts.

[ Sharpa ]

Our DR02 humanoid robot takes on the stairs with stable, controlled movement—steady steps, steady progress.

[ DEEP Robotics ]

Two words: structural minifridge. Or is it mini fridge...? Whatever, THREE words.

[ AgileX ]

Objects constructed by robots are ubiquitous. If you’ve used a car, household appliance, or smartphone today, you’ve used an object constructed at least in part by robots. The more products that manufacturers want to produce (and consumers want to consume) at lower costs, the more industrial robots will be needed.

There are over 4 million industrial robots in use worldwide, according to the International Federation of Robotics. And researchers predict that number will grow to over 16 million by 2030, as manufacturing rapidly increases. But what’s going to happen when they start breaking down? A new system designed by researchers at the Karlsruhe Institute of Technology (KIT), in Karlsruhe, Germany, can predict the defect in a broken product and disassemble it while protecting valuable parts from damage. To continue robotic development sustainably, the industry should prepare for the dismantling, recycling, and rebuilding of our robotic systems.

The system consists of a predictive algorithm that guesses how a product is broken, along with robotic manipulators that actually take the broken product apart. At every stage of the process, the system checks to see if the results align with its predictions, and updates its methods if necessary. For example, in the video below, the system begins by unscrewing a broken component. To simulate a stuck screw, the researcher replaces the screw. When the system observes the screw still in place, it switches to milling away material to remove the part.

Building a product with new parts is easy, says Jan Baumgärtner, one of the designers of the system. Each step is clearly outlined, and there are no expected deviations. But taking apart something that’s broken is unpredictable. “We can imagine 100 ways that something can go wrong.” And if you start taking something apart without knowing how it broke, you might have to undo part of your work when you find the problem. For example, if you have to unscrew 100 screws holding two parts together, but the last screw is stuck, you’ll have wasted time unscrewing all those screws when you should have used a different method to remove the part in the first place.

How to Take Apart a Product

KIT’s robotic disassembly system relies on a CAD model of the broken product and of each part, so it can see how the parts should behave and understand if anything is out of the ordinary. It also uses a mathematical model to predict the damage done to a broken part.

When you give the system a broken device and a CAD model, it first guesses how each part of the broken device should move. The axes each part can move along are called degrees of freedom (for example, a screw should rotate, but not move side to side). The disassembler nudges each part to see if it moves as expected. Based on how the part actually moves, it then uses the mathematical model to predict what went wrong with the part: A corroded part might move less than you think it should, a loose screw may move more, and a deformed part might have different degrees of freedom than expected.

At the beginning of disassembly, the system formulates a plan. It guesses what might be wrong with the device it’s taking apart, and then can change its guess based on observing each piece it takes apart. For example, if there was a screw loose in the part, that might be hard to guess from an initial photograph of the broken part. But when the system moves the screw, it will notice that it can move in more ways than a screw should move, and take that loose screw into account when deconstructing the device. You can also tell the disassembly system which parts are most important to salvage intact from a broken device, and it can adjust its strategy to preserve those specific parts.

The Automated Circular Economy

Baumgärtner’s motivation behind the design of the robotic disassembler is to help create a circular economy, where old devices are repaired instead of thrown away, reducing waste. “The big future is saving our planet,” he says.

Baumgärtner envisions scaling up this one system, composed of a few robotic arms, to have many robotic disassembler arms, each with different tools. These arms will specialize in a different part of the disassembly process so that an entire factory could use different robotic limbs to disassemble a wide range of products. Think of an industrial robot factory that creates cars, but instead is specialized to take them apart. Or, as he puts it, “as a giant robot with 100 arms.”

Ultimately, if this system works as intended, it would be a fully automated way of extracting a broken part from a system, replacing it, and rebuilding the device. Then the circular economy would really shine, as people replaced broken parts in old devices instead of buying new ones all the time. “That’s why we need to think about scaling this,” he says. “Because it means it becomes so cheap that it’s cheaper to repair this [electronic device] than to produce it. That’s the goal.”

This research was presented at the IEEE International Conference on Robotics and Automation (ICRA) 2026 in Vienna.

This story was updated 11 August 2026 to clarify that the disassembly system works for products in general, not only robots.