Valve Turning

Ladder Climbing

WISH US LUCK!!!

DRC-Hubo packed in its crate

How do I feel?  The closest feeling I can perhaps convey is the birth of one’s child, a wedding or graduation.  The time has come. One prepares the best he/she could and there comes a time when things are out of his/her control… the DRC will happen and the results will be what they are…

Drexel crew for Team DRC-Hubo (day before trailer goes to Miami)

Where did 15-months go?  I still remember October 24, 2012, the day DRC kicked-off with the world-wide webcast introducing the teams.  I still remember June 12, 2013, the day of DARPA’s Critical Design Review (CDR); we learned shortly after that our team passed and the Track B/C teams that would receive the Boston Dynamics Atlas robot.  Everything after that was a blur.  There was DRC Day in Atlanta (October 18, 2013) and learning we passed Qualifications (October 31, 2013).  Our team had a Dry-Run November 21-22, 2013 and “rehearsed” what we envisioned would happen in Miami – and discovered we still had some work to do!

How will we do in Miami?  On one hand, I’d like to say we already succeeded.  Getting into this Challenge, being very public about our progress, and demonstrated capabilities are all huge.  To participate against “giants” like NASA, Lockheed-Martin, Schaft (a University of Tokyo spin-off company), CMU and MIT, is a success.  To be a part of this level of “game” is a success…

But at heart, it’s about winning… to be among the Top 8 teams that DARPA selects in Miami to continue to the Finals in December 2014.  I can imagine the “agony” that my students would feel if we don’t make the Top 8.  I don’t want them to feel their “labors to be in vain and efforts to be futile” (Psalm 127:1).

That said, Dry-Run showed us that we can score at least 1 point in each event in the allotted time-limit.  DRC-Hubo can drive the vehicle (1 point); walk over rough-terrain (1-2 points); clear debris (1-point), open doors (1-2 points), break walls (1-point), climb the ladder (2-3 points); turn valves (2-3 points) and handle fire hoses (1-2 points).  Thus on a “bad day”, out of a top score of 24, we can score 10-points. On a “good day”, we can score 15-points.

Dec 13, 2013: Trailer with DRC-Hubo and Support Equipment hitting the road to Miami

We’ll see… this baby/wedding/graduation is coming in 7-days!  Wish us success.

Godspeed!

Why a DRC?  The DRC is about disaster response and humanitarian assistance. The tragic thing about the Fukushima Daiichi nuclear accident in 2011 was that if valves and switches could have been activated at the earliest stages of the meltdown, much of the nuclear catastrophe would have been prevented.  Radiation was too high for anyone to activate these valves and switches. Furthermore, wired and wireless connections were compromised.  So everyone asked where were the robots?  The sad thing is that existing robots were simply not capable of activating, let alone reaching, such valves and switches.  Thus the DRC attempts to have these technology gaps addressed, using the Challenge as a catalyst.  On the eve of the 12^th year since 9/11, disasters, whether accidental, natural, or intentional, is a stark reminder that mitigation concerns everyone.

October 23, 2012 Kickoff:  About 11-months ago, 30+ teams from various tracks were commissioned by DARPA to address the Challenge.  On July 11, 2013, down-selection whittled this number to 13 teams: seven were given the Boston Dynamics Atlas robot to use and six were allowed to continue using their own robot design.  I’m happy to say that our team (DRC-Hubo) was one of the six…

T minus 100 Days (T-100 Days): In about 15-weeks, team DRC-Hubo will start packing its trailer and driving 2-3 days from Philadelphia to Miami.  15-weeks may sound like a lot of time to some, but pragmatically it’s hard to get work done during Thanksgiving and the year-end Holidays; there are final exams at universities, holiday shopping and travel – and snowstorms in the Northeast.  With this mind, we continue to accelerate our efforts now to provide some buffer time later.

Verification Stage: The “teaser” video above is a result of a “verification” stage performed during the summer.  Called V&V (verification-and-validation), engineers assign a first phase to assess (verify) their design’s performance.  This phase is followed by a second one (validation) where performance is measured by non-designers.  The net effect is to ensure the design performs event specifications as engineered.  In these last 100 days, team DRC-Hubo is entering the validation phase.

I use the “smartphone and apps” as an analogy for our team’s approach.  The robot (nicknamed DiRC, pronounced “derk”) is like the “smartphone” and events are “apps”.  You want ladder-climbing, then launch the associated app, and DiRC executes the event.  You want vehicle driving, load that app, and our robot drives the utility vehicle.  Our team (led by Drexel) is composed of 8 universities with each school developing an “app” (8 of them) for DiRC provided by KAIST/Rainbow.

Unlike other teams that use a “not invented here” mindset, I feel our “app” analogy promotes “crowdsourcing” to let best practices emerge.  This helps accelerate design, reduce engineering burden and lessen risk.  Time will tell if our multi-university and international collaborative “invented together” approach will succeed… we have 100-days left to do so!  I’d love to see how the other teams are doing… and gauge if we’re on track, behind or ahead.

God speed!

It’s been a crazy-busy summer for Team DRC-Hubo and hence my blogging lapse (sorry!) so first let me explain the title. On July 11, 2013 DARPA gave their “green light” to 13 teams (6 from Track A and 7 Atlas-based ones) for competing in Miami this December – I’m pleased to say our Track A team is one of the six.  So, describing how our team got to this points might be a good way to kick-start this blog again.

Top 5 (oops, I mean 6) Track A teams Announced: On June 25, 2013 DARPA emailed our team saying that we passed our Critical Design Review (CDR).  However, DARPA had a press embargo in effect; the Track A teams that would go to Miami (called a down-select) would be announced July 11.  Towards this, DARPA requested Track A teams to create a 2-min video which they’d play at the news release.

Before going further, let me share what went on with CDR.  In early June 2013, DARPA visited the 7 original Track A teams to access their progress and to eliminate 2 teams.  This is called down-selection – forcing teams into a competitive Darwinian “survival-of-the-fittest” mode.   

The review was intense where we accelerated schedules and intensified efforts in preparation.  On June 12, 2013, we thus hosted DARPA’s “review” of our team at Drexel (see figure below).

June 12, 2013: Dr. Paul Oh (team DRC-Hubo Lead) kicking off CDR for DARPA (center front and seconds rows) with 20+ team mates (left and right sides of room) at Drexel.  Side note: Unbeknownst to DARPA we had Beta DRC-Hubo behind a black curtain (upper part of photo) to unveil at the right time

We described our engineering process involving testing-and-evaluation of Hubo and defining technical design requirements for the DRC.  We then:

• Unveiled our beta DRC-Hubo, nicknamed “DiRC” and pronounced “Dirk” (see below)
  
• Showed our full-scale mock-up test site, fabricated by a professional stage set designer featuring all the 8 events
  
• Our customized 24-foot Command-and-Control (C2) trailer which serves as our mobile “pit station” and will be driven to Miami from Philadelphia

DARPA Smiling: Beta DRC-Hubo unveiled showing range of motion (left) and steering (right)

A full-sized mock-up test site was fabricated and shown leading to smiles from DARPA folks (left). DRC-Hubo doing the valve-turning event (right)

24-foot trailer that team will drive to Miami (left), with interior customized into a command-and-control center (middle) and shown to DARPA at CDR (right)

Perhaps old news now, but we passed (see group photo below,  smiling but tired) – Go Team DRC-Hubo!

End of CDR day: Group photo with Team’s event leads and DARPA reviewers

Atlas Unveiling: So July 11 comes and the DARPA world-wide announcement is released.  Again, perhaps old news by now, but the announcement centered on Atlas, the Boston Dynamics humanoid that will compete in Miami.  Seven of these robots were to be given to the top 9 Track B and C teams.  Almost as a footnote, the press release mentioned the 6 (not 5) Track A teams also going forward to Miami and DARPA provided a short video of these Track A teams. 

A word about DARPA’s short video – it showed parts of our (old) notional concepts video that we created back in October 2012.  We provided DARPA with this clip showing some parts of DiRC but it wasn’t shown.  So here’s some unseen footage:

To be frank, I was a bit disappointed that the press announcement did not talk much about the Track A teams.  After all, we put so much effort into CDR.  More news about our (and other Track A teams) efforts would have been edifying.  But I understand – DARPA invested a lot of dollars into Atlas and hence their focus made sense.

Miami Semi-Final?  13+ teams (7 using Atlas, 6 Track A teams and an undisclosed number of unfunded teams) will meet in the Homestead Speedway in Miami December 20-21, 2013, to compete.  I call this a “semi-final” because DARPA said they’ll select the Top 8 performing teams.  These 8 teams will then have 1-year to prepare further and in December 2014 will compete to be the top team (and awarded the $2M prize).

I believe our chances of making the Final 8 are strong.  Team DRC-Hubo just concluded its 10-week “summer boot camp”.  Our 8 partners all gathered at Drexel, sleeping and dining together in dorms, coding together in our “base camp”, and verifying their approaches in the mock-up.  It was intense but it built up both “DiRC” and team morale into a long-lasting memory.  It’s not just about building robots, but about building friendships and building the knowledge base – and most importantly, building for tomorrow’s disaster-response.

God speed!

Slide from Kickoff Meeting on Robot Sensor Head

In less than 2-months, software teams will compete in the Virtual DRC.  The top teams will get their own Boston Dynamics Petman/Atlas humanoid!

Big Question: So what comes with the Petman/Atlas robot?  And more specifically, what are the sensors?

Possible Answer: At the October DRC Kickoff Meeting, I snapped a photo of the slide describing the Robot Sensor Head. Apparently a company called Carnegie Robots will provide a MultiSense-SL with the Petman/Atlas.  The DRC website has a datasheet describing this head.  It isn’t very detailed but from looking at the photo, my “best guess” is that the head employs a Hokuyo UTM-30LX-EW LIDAR sensor to provide range data.  I asked some colleagues for their “best guess” and they agree.

Why the Hokuyo UTM-30LX-EW?  First off, a DRC robot needs a suitable sensor to gauge range, and LIDAR (Laser Imaging Detection and Ranging) is a viable choice.  Second, the sensor should be light-weight and operate outdoors.  Looking at Hokuyo’s product line, a photo of the UTM-30LX-EW matches the ones in the DARPA slide and the MultiSense-SL datasheet.

Specifications for the UTM-30LX-EW: (See datasheet)

• About $6000
• Uses Multi-Echo Detection for outdoor usage
• Range of distances: 0.1 to 30 meters
• Accuracy: around 10 mm
• 270 degree field-of-view
• Mass: 210 grams
• Interface connection: ethernet
• Dimensions: Width (63 mm) by Depth (66 mm) by Height (87 mm)

One might know that Willow Garage’s PR2 robot employs a similar LIDAR sensor called the Hokuyo UTM-30LX.  It’s similar to the UTM-30LX-EW but employs USB interfacing and doesn’t have multi-echo detection (and hence not suited for outdoor use).

LIDAR output using the Hokuyo UTM-30LX-EW (right) measuring our surrogate DRC vehicle (left)

Sample Trials Using the Hokuyo UTM-30LX-EW: Our lab acquired one of these sensors.  In the above photo, the sensor captured range data (right) of a golf cart (left).  Here’s a video using the sensor to segment different areas of the vehicle.

Tutorial on the Hokuyo UTM-30LX-EW: A quick Google search doesn’t show much info beyond performance specifications.  So, I had my student (Mr. Karthikeyan Yuvraj) write a “Getting Started” tutorial. He’s working with our DRC partner Prof. Christopher Rasmussen from the University of Delaware.  Below is a photo of his prototype sensor.  One can see the Hokuyo UTM-30LX-EW mounted below the cameras (binocular lens).

Prototyping a sensor head for Team DRC-Hubo

The tutorial assumes knowledge of Linux (Ubuntu) and ROS (Robot Operating System).  The tutorial uses open-source tools like the 3D visualization tool RVIZ.

Summary and Impact to DRC: The Hokuyo UTM-30LX-EW is a relatively affordable sensor that can fit on a DRC robot without taking up much space or consuming much power.  It can operate outdoors and hence meets DARPA’s requirement that all 8-events of the DRC will be held outdoors.

One issue is that the Hokuyo UTM-30LX-EW is only effective in sensing objects between 0.1 to 30 meters. In other words, the robot will be “far-sighted“; it may be able to “see” objects far away (as is the case for vehicle-driving), but not those that are close (as may be the case with handling objects like valves, door-handles, hose-connectors and ladder-rungs).  The net effect is that we must give DRC-Hubo “bifocals” – and hence add sensors (like cameras) for “near-sightedness“.  Alas, our sensor head prototyping continues – stay tuned for more product reviews and tutorials…

Godspeed!

The DRC-Sim: The Boston Dynamics robot (right) towers over the Robotis Darwin-OP (middle) and is head-and-shoulders taller than the our Hubo (left)

Yes, that’s correct – that’s our team’s robot Hubo on the left, the off-the-shelf Darwin-OP from Robotis in the center, and the Boston Dynamics Petman/Atlas on the right – all in the DARPA Robotics Challenge Simulator (DRC-Sim).  In the background is a computer model of our golf cart which serves as our surrogate test vehicle.  Links to tutorials are given at the end of the blog.  Here’s a YouTube video of the above screenshot.

Why Care? The DRC-Sim and Virtual DRC is an example of crowdsourcing – generating solutions to complex problems by leveraging the efforts of many diverse people.  The DRC-Sim and the chance to be awarded a real Boston Dynamics robot provide both tools and incentive to get many solutions.  From what I heard, over 100 self-funded teams registered for Track C to compete in the Virtual DRC (V-DRC) this June.  It will be exciting to see what happens!  With about 6-weeks until the V-DRC qualifiers in May, we’re beginning to see some cool results from Tracks B and C teams.  Check out Dr. Jerry Pratt’s DARPA-funded Track B team YouTube videos!

Do You Need a PhD to Use DRC-Sim?  Not exactly – but remember, the DRC is what is called “DARPA-hard” – something that is very difficult but not impossible to achieve.  Past examples of DARPA-hard include landing man on the moon and radar-defeating stealth aircraft – not impossible tasks but very difficult to accomplish.  The fact that there’s a $2M prize for the winner of the DRC suggests the 8-events of the Challenge are indeed DARPA-hard.  Otherwise, wouldn’t anyone and everyone enter?

Frankly, those who are computer-literate can get DRC-Sim installed and running and those with Linux/Ubuntu experience can do so even faster.  The difficult part is what to do afterwards!  You have to write algorithms that command the robot to walk, enter a vehicle, climb a ladder, operate hand-tools and turn valves.  Understanding biped walking gaits, dexterous manipulation and motion planning often demand graduate-level robotics courses.  Even then, what works in simulation does not always work in the real world; things like friction, the interactions of the feet and the floor (e.g. walking on pavement is different than on sand), and dealing with lighting (e.g. shadows, glare, etc) all make the DRC “DARPA-hard”.

So Why Darwin-OP?  I see the DRC as a teaching moment for building a community.  The U.S. lags far behind Asia and Europe in humanoids.  As such, the DRC is an opportunity for the U.S. to re-build momentum in this area of robotics and join other countries to showcase capabilities and potential.  However, working with full-sized walking humanoids requires a lot of resources.  In other words, the “barrier-to-entry” is high not only for general people but for research institutions like agencies, industry and academic labs.

Mini-DRC is a project to “teach” my less-experienced students about humanoids. Josh
Geating at Drexel custom-made a “car” for Darwin-OP to drive so these students can mimic the real DRC’s Event 1 (utility car driving).

Let’s Have a Mini-DRC!  As a robotics professor (and a Track A team leader), I want students to be able to “participate” in the DRC – especially if they don’t have access or experience with real humanoids.  I’m actually forming a Mini-DRC with students where the events are scaled to Darwin-OP’s size.   For example, Drexel student Josh Geating custom-made a Darwin-OP sized vehicle (see figure above).  See YouTube video of Darwin-OP “driving” a custom-made “car”

By having students use the Darwin-OP model, I hope they will appreciate the DRC-Sim and the disconnect between simulation and “reality”.

Tutorials: As such, I assigned Drexel engineering student Bharadwaj Ramesh who had no Linux experience go through the Tutorials on the DRC website.  He authored his own tutorials (including how to install Ubuntu) and finally ended his assignment with importing Hubo and Darwin-OP models.  See YouTube videos of Hubo and Darwin-OP in DRC-Sim.

As for Team DRC-Hubo, we have our own simulator (called OpenHubo) based on OpenRAVE – a well-known motion-planning simulator.  If you’re interested, you can download the Hubo CAD models from Git Hub and try developing your own algorithms in either DRC-Sim or OpenHubo.

More tutorials will come.  Part 2 will show how to make Darwin-OP walk in the DRC-Sim.  Stayed tuned.

Please let us know what you think!  Especially about Mini-DRC, or using Darwin-OP and/or Hubo in DRC-Sim or OpenHubo!

Godspeed!

Photo adapted from a Forbes April 2006 article showing the Boeing 787 was created “globally”

At first, I thought his statement was just marketing rhetoric.  But I made a similar mistake that year again.  In an audience of executives in Seoul, I said “Samsung is a Korean company”.  Their VP corrected me saying that Samsung isn’t Korean but is a “Multinational” company.

Was this just rhetoric?  Mind you, this was 2006 – about the time “The World Is Flat” by Thomas Friedman became a best-selling book and before the word “globalization” became part of everyday language.

I grew up associating brands with countries; Rolex is from Switzerland, BMW is from Germany and Sony is from Japan.  However, like many things today, associations are blurry.  Are Hyundai cars assembled in Alabama still Korean?  Are iPhones manufactured in China still American?  Sure, one can debate nationalities of products, but one can also appreciate that the “borders” are blurry and less defined that in years past.

Hubo is “Global” too: Let me begin by saying Hubo was conceived, designed and developed by Prof. Jun Ho Oh at KAIST (Korea’s leading science and engineering university).  Since its debut in 2000, each generation of this humanoid captured international headlines.  In some senses, Hubo is an icon of Korea’s robotics program.  In 2008, my lab acquired a Hubo from KAIST through the US National Science Foundation’s PIRE program.

Like Boeing’s 787, Hubo is a result of a US-Korea partnership

Like computers, Hubo needs application software to be useful.  The goal of our grant was to marry KAIST’s world-class robot hardware designers with America’s world-leading robot software researchers.  This global marriage endeavored to catalyze humanoids by giving the robot the (machine) intelligence to physically do useful tasks.  Without “American” software, Hubo would simply be a mechanical puppet and without “Korean” hardware, the “intelligence” could not be applied… Hence, I argue (and I think my colleagues agree), Hubo is “global” too.  The figure shows the different areas of Hubo partners are working on.

US Students at Risk: The National Academies report that less than 3% of American science and engineering students have any significant (non-tourist) experiences abroad. By contrast, over 50% of European and Asian students leave their continents for 6+ months working in labs overseas.  These aren’t back-packing vacations or globe-trots.  Rather, European and Asian students are learning cross-cultural skills and developing international networks to work in “global” companies.  These students will be the designers of tomorrow’s consumer electronics, gaming consoles, automobiles and airplanes.  They will fuel their country’s economies with new products, multinational markets and global services…  Can American students compete given such statistics and mindsets?

Hubo – Teaching US Students Skills for International Collaboration:  Unlike perhaps many liberal arts students, American computer science and engineering majors see little need to go overseas to learn.  After all, with the internet can’t one simply collaborate over email and Skype?  Also, isn’t America the “best” technologically – so why go abroad?

I asked the same question when I started our PIRE project.  I doubted that my students would study 3-terms of Korean language, live in Korea for 6-months working full-time in the KAIST Hubo Lab, and engage in a culture very different from America.

To my surprise, many of my students were willing to leave their “comfort zone”.  The opportunity to get another key-hole into life, double their perspective by seeing the world outside of America, and to work with a world-class robot was too good to pass up.  I’m happy to say that from 2008 to 2012, I’ve sent over a dozen American students to Korea – you can see their blogs here.

For many of these students, Korea was their first trip outside of the US (and for some, even outside of Philadelphia).  Some never flew on an airplane before.  Some never had a passport – nor did their parents.  Some were nervous (doesn’t the world hate Americans?).  Some would miss Thanksgivings, family reunions and friends’ birthdays.

But they all returned intrepid world-travelers!  They gained confidence and boldness.  They learned to adapt to new cultures and relationships.  They self-actualized.  Moreover, they learned how to collaborate with non-Americans developing a very complex robot – skills they can translate when they join the “Boeings” of tomorrow to create “global” products.

After DRC, perhaps tomorrow’s Hubo will be a product of many nations – just like a 787

DRC is Global Too:  As you probably know now, the DRC was motivated by the limitations of the Fukushima disaster response.  Fukushima had world-wide impact, not only with radioactive spread but with disrupted global supply chains.  As such, DRC-Hubo has even more significance than simply bringing Korean and American roboticists together.  DRC-Hubo provides both “learning” and “teaching” moments ranging from technology, international collaboration and designing products for the greater good of all mankind.

With that, Hubo may have originated from Korea but today is a result of a US-Korea collaboration for tomorrow’s global needs.

Godspeed and Open to your comments!

Site Visit Results: I’m happy to say DARPA is pleased!  We gave 4 live demos of Hubo entering the vehicle, opening a door, turning a valve and attaching a hose (i.e. Events 1, 4, 7 and 8).  See photo below:

Hubo doing Event 1 (top left), Event 4 (top right), Event 7 (bottom left) and Event 8 (bottom right)

Our team mates at Ohio State (OSU), Purdue and Georgia Tech (GT) continue to amaze us with what they have accomplished in their events.  DARPA saw their videos of OSU-Hubo walking over rough-terrain, Purdue-Hubo climbing a ladder and GT-Hubo breaking a wall.  All in all, DARPA looked happy with our progress and our plans moving forward.

We Even Published: To my best knowledge, we’re the only team already demonstrating these events.  So, we wrote technical papers on each event, which were peer-reviewed and accepted for publication (yippee!) in the 2013 IEEE International Conference on Technologies for Practical Robot Applications (TEPRA).  Each paper details our approach to all 8 events, illustrates simulations and discusses the experimental results.  We expect that IEEE will publish are papers in late April.  Once they are published, visit us again to see videos and more photos.

Why Publish? You may wonder why we are publishing.  Wouldn’t such disclosure tip off our competitors and undermine our chances for the $2M prize?  Perhaps this is true.  But moreover, we are a team composed of universities.  Universities exist to create, archive, and disseminate knowledge – for the greater good.  This is done through both teaching and research, in classrooms and labs. Our team believes that advancing the state-of-the-art of disaster-response robots demands an open review and scientific verification and validation by the robotics expert community.   Our belief in the “greater good” compels us to publish – and together with those beyond our team, help advance robotics.

Let the Site Visits Begin!

Why this blog? The DRC is a defining moment in robotics and will fundamentally change our views and interactions with robots.  I thus want to chronicle our experiences and capture “history-in-the-making”.   By doing so, I hope to engage others to join and cheer us in this journey.

Who this blog is for?  I identify the following 4 audiences:

• Those interested in the DARPA Robotics Challenge (including our competitors!)
• Roboticists from academia, industry and research labs (i.e. my colleagues!)
• Robot enthusiasts
• My family and friends wondering what I’m doing!

What will this blog cover?  As “history” unfolds, I will journal aspects of the DRC, our trials and tribulations (and hopefully joys too!) and occasionally, my views of robotics.   These views reflect my role as team lead, but also as an educator and robotics researcher.

How and When?  I’m not a professional blogger and appreciate that many blogs go stale or die.  Nonetheless I will try to blog at least weekly.  I will also experiment with hosting video interviews and using social media tools.

Next steps: Please join us on social media sites like Google+, Twitter, YouTube and Facebook.  You can find our links on this page or http://www.drc-hubo.com.

My next blog will be on my conviction that within 7-years (i.e. 2020), disaster response robots will be part of our daily lives as seen in this notional video.

Thanks, Godspeed and God Bless!

DRC-Hubo will be a version with hardware retrofits and algorithms for autonomy. 7 Hubos exist at Drexel University (lead PI) and are immediately available for the DRC.

Our Vision of Disaster Response (Year 2020)

Share:

]]> http://www.blog.drc-hubo.com/?feed=rss2&p=19 0