Artificial Intelligence and Robotics

RoboCup Rescue 2009 photos

2009-07-14T01:44:00.000-07:00

Amir, one of the participants in this year's RoboCup Rescue competition was kind enough to let me know of a large collection of photos from the competition that he would like to share with all of us. For context, the rescue competition is designed to stimulate research in building autonomous or semi-autonomous mobile robots that can assist in rescue operations in disaster zones, e.g., finding survivors in a collapsed building after a major earthquake. The competition has been taking place annually for more than a decade and I can tell you that the rescue robots are becoming more advanced every year. I am very optimistic that in another 10 years such robots will become a rescuer's best friend helping save many lives.

With Amir's permission I have included a couple of the photos at the bottom of this post, but if you want to see the entire collection of more than 100 high resolution photos, go here.

Thank you Amir and I hope your team was successful at the competition.

Wiimote control of a 15-tonne machine

2009-07-12T01:28:00.000-07:00

If there is anything Australians know how to do well, it is mining. Transmin engineers had some fun building a steel monster controlled using a Wiimote. It is not something that you will be able to buy for home use (not that you would ever want one of these machines for home use) but it is a fun little project attached to a very serious project of building a massive piece of equipment for mining. Watch the video and share in the fun.

Thanks Adrian for the heads up.

Multi Autonomous Ground-robotic International Challenge (MAGIC 2010)

2009-07-08T03:41:00.000-07:00

It seems like yesterday when CMU won the DARPA Urban Challenge and Stanford the DARPA Grand Challenge. And yet, it was nearly 1.5 years ago when the most recent of the two concluded. A similar challenge event also took place in the UK and Google still has an ongoing Lunar X Prize. The Multi Autonomous Ground-robotic International Challenge (MAGIC 2010) cosponsored by the Defence Science & Technology Organisation (DSTO) in Australia and the Research Development & Engineering Command (RDECOM) in USA is a new event that was announced at the begining of this month.

MAGIC is different than the other challenges because it no longer focuses on the development of a single intelligent vehicle but an entire team of cooperating intelligent vehicles. The goal of this multi-agent team will be to perform "intelligence, surveillance and reconnaissance mission in a dynamic urban environment."

Creating a team of cooperating agents is a much more difficult task than creating a single agent; these agents have to be able to share information and workout cooperative plans that take advantage of all information available to all agents which may be difficult to communicate in certain circumstances.

On the other hand, it can be far more efficient to use multiple vehicles to complete a particular task. If the task can be broken down into smaller tasks that different vehicles can complete and then merge the solutions, e.g., cooperative mapping of a large space can be achieved twice as fast by two vehicles compared to just one.

MAGIC participants will have to complete the following tasks:

(i) Accurately and completely explore and map the challenge area; (ii) correctly locate, classify and recognise all simulated threats; and (iii) complete all phases within 3.5 hours.

There are some prize money for the winners and a small amount of funding for select participants. More specifically, 5 teams will receive funding of $100,000 each. The top 3 teams will receive $750,000, $250,000, and $100,000 respectively. The final event is scheduled to take place during the week of November 8, 2010, somewhere in South Australia.

I'm curious to see what kind of multi-vehicle teams robotics researchers design to tackle this challenge.

How does the Willow Garage Personal Robot work? Watch the video to find out.

2009-07-07T02:12:00.000-07:00

A bit less than a month ago, Willow Garage demonstrated their Personal Robot 2 achieving the second milestone on their roadmap to building an open source, dexterous, mobile robot. They successfully demonstrated the robot navigating an office-like environment, opening and going through doors, and finding outlets to recharge its batteries. You can watch the video of the demonstration in our previous post here.

A few days ago, the Willow Garage engineers posted a new video explaining some details about how their robot actually works. In the video below, Sachin Chitta and Melonee Wise explain how the robot uses occupancy grid maps to navigate dynamic environments, uses vision and laser information to detect door handles and power outlets, and, finally, plan its actions to satisfy its goals.

Watching the video, I have to admit that it must take some serious effort to calibrate all these sensors. If the sensors are not calibrated and the relative poses among them are not known, it will be very difficult to fuse the data from them into a single, accurate estimate of the world's state. It would then be very difficult for the robot to act correctly in the presence of noisy sensors and actuators. In other words, I think achieving this milestone is not a trivial task and the Willow Garage team is certainly doing a great job.

The video explanation of how PR2 achieved Milestone 2 follows.

Flapping-wing Nano Air Vehicle (NAV) from AeroVironment

2009-07-03T03:13:00.000-07:00

According to a recent press release, DARPA has agreed to continue financing a research program for creating a small flapping-wing flying robot. AeroVironment are the happy recipients of 2.1 million dollars to continue NAV's development after achieving several breakthroughs during the now completed phase I of the project; phase I started in 2007 and lasted for 2 years while phase II is expected to continue until the summer of 2010. Specifically and as you can see from the below video, AeroVironment engineers have successfully built a small flapping-wing robot capable of hovering and flying in all directions under remote control.

The company plans to develop a robot that weighs no more than 10 grams and can be controlled from up to 1 mile away with a top speed of 10 meters per second. Obviously, there are numerous military applications for such robots including surveillance, reconnaissance, and even delivery of deadly payload with high precision.

For another nice high speed video of a flapping wing micro-robot, check out our previous post on the Butterfly Ornithopter.

Cute robot makes coffee

2009-06-30T15:06:00.000-07:00

HINA, one of the cutest robots to have ever been created is shown in the carefully edited video below making a cup of coffee for her owner, or better yet, her friend. Although the robot does do all the work as shown, it definitely did not do any of it under its own will or on a single try. Just think of the video like a well choreographed kung-fu fight in a movie. Enjoy!

More information about the robot here (website in foreign language which I am guessing is Japanese.)

Netflix prize claimed by international research team

2009-06-27T02:06:00.000-07:00

In the span of 3 years, some 40,000 teams from around the world took up Netflix's challenge of improving movie recommendations by 10% over the company's Cinematch engine. And just a day ago, one of these teams has claimed the top prize and the $1,000,000 that go with it. This team consists of the 2007 and 2008 progress prize winning team BellKor from the USA, Austria's Big Chaos team, and Canada's Pragmatic Theory team. Until recently, the teams were competing against each other unable to reach the 10% improvement threshold but they put an end to the competition when they joined forces.

Other teams now have 30 days to submit their solutions and outdo the top team for a chance to claim the top prize for themselves; keep an eye on the official leaderboard here. Can others do it? Well, you never know so it will be an interesting 30 days.

I have to admit that a couple of years ago I really didn't think that this day would come. I was certain that the 10% improvement had been selected by Netflix to prevent people from ever claiming the million dollar prize. I guess, I was wrong after all.

That said, I am currently reading Don Tapscott and Anthony D. Williams' best seller Wikinomics: How Mass Collaboration Changes Everything and I think the Netflix competition would make for a great story for this book. It is yet another example of how computers and the Internet allow thousands of people across the globe to collaborate effectively and efficiently to solve problems at a fraction of the cost of a single corporation doing the work in house. I don't know how much money in Research and Development Netflix will save from this competition but I suspect they stand to make a lot of money from the 10% improvement in recommendation accuracy. They will certainly make back the prize money in no time (if they haven't already considering that improvements over 7% have been achieved by many teams over the course of the last 2 years.)

Kudos to the teams for their achievement and kudos to Netflix for taking a chance most corporations would never dare take.

AutoStitch: Panoramic image creation for the iPhone

2009-06-23T02:58:00.000-07:00

The iPhone app store may be home for some 50,000 applications but few are actually worth the download, free of charge or not. One of these few applications that are a must have for every iPhone owner is AutoStitch by Cloudburst Research Inc.

This nifty application can create panoramic or wide-angle images from any arrangement of photos and minimal user guidance. The latter two features are what makes AutoStitch such a great application. Any arrangement of photos means that you don't have to take your photos in sequence for the stitching to work; the application can figure out the overlap between images on its own by the use of state-of-the-art feature-based image matching techniques. And minimal user guidance means that you only have to point the application to the directory where all your photos are stored and it will gladly stitch together all those photos that belong to the same scene. The photo at the top of this post is an example panorama created using this application (the image is copyright Cloudburst Research Inc.)

If you recall, I wrote in the past about image stitching software including the high resolution Gigapan system, the desktop version of AutoStitch, and the open source autopano-sift. Microsoft's Photosynth maybe the one related application that is best known to you. Similar software is available today in several commercial image processing software packages. Interestingly, the founders of Cloudburst are also the first to publish the method that makes Autostitch and all these other panorama creation software work so well.

Learn more about the iPhone Autostich application at the company's website here or you can buy the application from the App Store for $1.99 here.

New book: Computer Vision Algorithms and Applications

2009-06-19T17:24:00.000-07:00

There is a new computer vision book in the works but a well know researcher Richard Szeliski of Microsoft Research. If you don't spend much time reading conference proceedings and journal articles you may not have heard of the author before. However, you know him indirectly from some of his work that Microsoft has started publicizing in the last few years. Richard Szeliski is one of the main people behind Photosynth, the software that allows users to create stunning panoramic images from collections of digital photos without much hassle.

The new book is based on the lectures of a computer vision course that Szeliski has taught with some of his colleagues at the University of Washington. The book starts with a description of some basic concepts on image formation and processing. It then continues to cover a large number of advanced topics in feature detection and matching, segmentation, calibration, structure from motion, image stitching, computational photography, stereo, recognition, and image-based rendering. The book presents recently published work by a variety of computer vision researchers from across the globe so many of the chapters describe state-of-the-art methods.

That said, Szelinksi's new book is not finished yet so don't rush to the bookstore to buy it. While still working on the project, the author makes the most current draft available online for anyone to download and read. He seeks the community's feedback in making this a book worth having. I have read parts of it and it looks like it is going to be a great book when finished. You can download the latest draft here.

Willow Garage robot continues to advance

2009-06-17T03:30:00.000-07:00

Earlier this month, Willow Garage's Personal Robot 2 (PR2) robot successfully passed the second project milestone navigating an office environment, opening doors, and plugging in the electrical supply to recharge its batteries. The first milestone the robot successfully passed last December (if my memory serves me well) was about navigating around an office environment but without the requirement of opening doors.

According to the company, the robot navigated autonomously for nearly one hour passing through half open doors and locating power sockets while realizing a locked door was not passable and moving on to another one. Apparently, the robot's batteries allow it 3 to 6 hours of autonomous operation. PR2 has two compliant arms with 7 DOF each.

Willow Garage was founded only in 2006 and the robot they have constructed so far has some very impressive capabilities. Sure, it is no ASIMO but the team has much less money to work with that the Honda team. Let us also not forget that the same group is developing an open source operating system for robotics, support the open source player/stage robot control architecture, and continues to update the open source computer vision library OpenCV.

The video below shows PR2 navigating the Willow Garage offices and doing its thing.

PS: I wonder what the 3rd milestone for the project is supposed to be.

Robe:Do makes affordable robots for fun and games

2009-06-14T04:29:00.000-07:00

The market of affordable robots for hobbyists has definitely flourished in the last few years as electronics components and open source software have made such creations possible. Robe:Do Robotics is a small mom and pop operation out of Colorado that builds and sells a small range of such small and affordable robots that look like a lot of fun.

The company is owned by William and Esmaa Self who hand build every robot they ship to customers. For now, they sell 3 robots called the Coppa, Silvio and Three. The first two have 4 wheels and are designed for carrying large payloads while the last one which is also their latest offering is designed more for speed than heavy lifting. Three can carry a laptop up to 6 pounds heavy. Robe:Do suggests that the $439 robot be paired with an equally cheap netbook computer for many days of fun activities. The company provides high-level libraries for controlling the robots making it easy to get one running around avoiding obstacles with just a few lines of code.

In case you are wondering what can one really do with robots of such modest specifications, take a look at the below video of a Coppa robot receiving a command from an iPod Touch and then preparing and carrying popcorn just in time for the movie.

Pleo to be manufactured once again

2009-06-11T01:18:00.000-07:00

News the last couple of days are that the Hong Kong based Jetta Company Limited has acquired the Pleo IP from Ugobe and they have promised to start manufacturing the robot toy once again in the near future. The company's press release was not any more specific about exactly how and when Pleo will be brought back to life.

Back in April when Ugobe filed for Chapter 7 bankruptcy, I mentioned a number of reasons as to why Pleo failed in the market. In summary, the product was overpriced and overhyped; the cost of 350$ was too much for a luxury toy during an major economic downturn. Ugobe's IP was expected to sell for a good amount of money and so it is no surprise that a large and successful company such as Jetta eventually purchased it. I guess I should mention that Jetta was the original manufacturer of Pleo working under contract with Ugobe during the dinosaur's brief existence on Earth.

I am curious to see what Jetta will do with Pleo. Are they going to manufacture and sell Pleo exactly as it was last or are they actually going to upgrade it in some ways? Is Caleb Chung going to be involved in future development? This latter point may be important for the continued survival of this robotic toy. At the end of the day, I believe that much of the media coverage and the hype surrounding Pleo had lots to do with the fact that it was the brainchild of Caleb famous for being the creator of the Furby toy. Without him and without any major improvements of the current product, will anyone care for Pleo?

We will know in a few months what Jetta has in store for Pleo. In the meantime, owners of Pleo can feel relieved that there is a good chance new parts such as batteries will continue to be available and their expensive toy will not end up becoming nothing more than a Jurassic paper weight.

Reach: Robot video

2009-05-29T12:24:00.000-07:00

While I am away for another week and posting on this blog is sparse, enjoy the below video of a cute little robot reaching for its dream. The video called "REACH" is Luke Randall's entry in the Cannes Short Film Corner competition.

The autonomous city explorer robot

2009-05-17T06:11:00.001-07:00

It is not every day that you see robots traveling long distances around a city without GPS and with the help of perfect strangers. A couple of weeks ago, I wrote about the tweenbots project which used the kindness of strangers to guide the robot to its destination. But that project used a very simple kind of robot not capable of much more than moving on a straight line.

Researchers at the Technical University of Munich have been developing a city exploring robot that is capable of interacting with strangers using speech, hand gestures and a touch screen to get information from them. The Autonomous City Explorer (ACE) project has successfully demonstrated the robot navigating 1.5Km around Munich.

ACE is a very complex robot using a number of advanced sensors and algorithms to sense the surrounding environment, move without colliding with obstacles and, of course, interacting with other pedestrians who stop to help the robot find its way. The robot which comes equipped with both stereo vision and laser sensors stands 178cm tall and weighs 160Kgrs. A total of 3 onboard computers provide the necessary processing power for navigation and interaction. Lastly, the robot can move at a maximum speed of 1.4m/sec and slopes of up to 6 degrees but obviously at a much slower speed.

ACE navigates by constructing topological maps of its environment as it moves along. It uses specially designed modules to detect the traversability of regions used for path planning and following. It is capable of detecting people using skin color detection and tracking them as they move. The interaction interface using speech and gestures can be a bit crude as it requires people to stand in a way that is easy to interpret but it is a step ahead anything else out there.

The video below describes the system details and shows the robot in action interacting with other pedestrians and safely navigating in an urban environment.

Hawk: The Canadian humanoid robot

2009-05-09T15:23:00.000-07:00

A small Canadian robotics company that has been around for nearly a decade but we hardly ever hear any news about recently started selling their own version of a humanoid robot with dual arms and a wheeled base. Hawk is Dr. Robot’s research, education, and entertainment robot.

The robot stands 1.4 meters tall and comes loaded with a number of sensors including vision, sonar, and laser (an optional upgrade to the base system.) Hawk resembles a human only in its upper body which consists of a torso complete with two arms (6 degrees of freedom each) and a movable head with stereo vision. A small touch screen on the robot’s chest allows for a more reliable method for commanding the robot.

The robot’s base is an i90 wheeled base made by the same company. Dr. Robot claims that Hawk can operate from as little as 2 hours to as long as 8 hours before recharging. The longer time can be achieved after an upgrade to the robot’s battery pack. Finally, Hawk can lift weights of 0.8-1.0 Kgrs which should be enough to carry small items around the house.

It is not clear to me how autonomous the robot can be when it comes to performing complex tasks but it comes with cross-platform software allowing users to tele-operate Hawk. The company claims that the robot is capable of simultaneous localization and mapping which should allow it a good degree of mobility around a house. For example, its navigation capabilities including visual locating a charging station and autonomously docking for recharging its batteries.

The video below shows the robot Hawk performing part of a serving task opening a bottle of water and pouring it into two different glasses.

Wolfram Alpha demo video

2009-05-03T14:45:00.001-07:00

Stephen Wolfram's new search engine Wolfram Alpha is scheduled to go live this month. It is not online yet, by Stephen has been going around showing off some of its capabilities to academics. He recently did a long presentation at Harvard University which if anything increased the hype surrounding Wolfram's computational knowledge engine. The 10-minute video below shows a small part of the presentation including screen shots of the engine's responses to a few of Stephen's queries. It definitely looks like a very interesting service and I am looking forward to using it. I just have to figure out what would be the right questions to ask in order to really take full advantage of the machinery behind the interface. Hmmmm...

SimplySim: A very impressive 3D simulation environment

2009-04-30T14:50:00.000-07:00

I was recently told about SimplySim a new French startup that specializes in creating realistic 3D simulations. The small company was established a year ago with the hope of offering the best simulation software and related services to the robotics community.

SimplySim simulations are not only visually pleasing but also physically correct. The software integrates nicely with the Microsoft Robotics Developer Studio providing complete and detailed 3D environments and sensor models such as the sonar one recently released. In fact, any 3D environment created using one of the many tools included in Simply Suite can be exported and used in a robotics simulation within MRDS. The SimplySim team has recently released an MRDS simulator for the Lego NXT Mindstorms which is free to download from the SimplySim website.

The simulated robot is composed of a differential drive, an ultrasonic sensor, a light intensity sensor and a bumper. You can also control the claws. It moves in a fully physical environment with several objects.

The team is also working closely with Microsoft to support the Imagine Cup event and more specifically the Mars Challenge competition. This challenge requires students to develop software that would guide a NASA rover to explore near a crater on the surface of Mars performing science experiments classifying different rock types while looking for evidence for the existence of water on the Red planet.

It is difficult to appreciate how good the 3D simulations are from a textual description thus I present below a video showing off some of the main features of the product including the above mentioned NXT simulator package.

Pleo goes the way of the dodo bird

2009-04-25T03:11:00.000-07:00

I was finally proven wrong in my criticism of the robot dinosaur toy Pleo and its maker Ugobe. It turns out that Ugobe indeed built an exact replica of a dinosaur, a real life form; so exact in fact, that Pleo went extinct just like the real dinosaurs. In retrospect, Pleo's demise can be attributed to a number of reasons that I will summarize shortly. In the meantime, just in case that you haven't heard it yet, Ugobe filed for Chapter 7 bankruptcy a few days ago. The company had to let go all of its employees (unfortunately) and is currently in the process of liquidating its assets.

I said since the introduction of Pleo in 2006 that the product was over hyped. It definitely created a niche following among some early technology adopters but it never hit the big time even though it was making huge headlines in all major (and not) publications and TV shows. It didn't take long, however, before people started to complain that their robots weren't capable of doing more than walk around slowly and make random sounds; needless to say, the learning that a Pleo was supposed to be capable of over time did not live up to expectation. Even so, Ugobe was successful in selling about 100,000 of these toys that originally retailed for $350. Recently Amazon reduced Pleo's price to $250 but it was too little too late fro the ambitious robotic toy.

To a large degree, Ugobe fell victim to the recent economic downturn that has hurt many businesses. Their offering was too expensive for a toy in an economic climate that will be remembered for huge job losses and a shut credit market. Failing to improve their original offering of 2006 didn't help Ugobe either. There were rumors of a newer, better version of Pleo that never materialized or maybe it never did because bankruptcy hit too soon. The personality upgrades made available via the official website also fell short of expectations.

The company says that they tried to sell or merge with a bigger company in an effort to continue their work on Pleo but it just was not meant to be. Ugobe's IP which includes a number of patents is expected to fetch a decent amount of money when put on sale.

In general, the consumer robotics market is currently under a lot of stress because of the ongoing debt and credit crisis. The truth is that there exist no robots in the market that perform a useful task at a price that would justify a money strapped consumer to purchase it. I would not be surprised if sales of other toy and consumer robots such as those manufactured by WowWee and iRobot have also gone down in volume in the last few months and probably remain lower than usual for the rest of the year. Of course, both WowWee and iRobot are large enough companies to survive the current economic conditions but a small company such as Ugobe had little hope of doing the same.

I just hope that the Ugobe employees who find themselves unemployed in such difficult times are able to find new work quickly.

Tweenbots: Human-dependent robots navigating a big city

2009-04-11T05:02:00.000-07:00

Tweenbots are cute little robots with few skills of their own other than being able to slowly move forward. At 10 inches high, tweenbots have a hard time navigating even the smallest of obstacles needing lots of help from humans passing by.

The tiny robots are the creation of Kacie Kinzer, an arts student interested in studying something I don't understand, "...finding ways to provided more textured and meaningful experiences through technology--particularly in relation to the transmission of narrative." Regardless, this Tisch School of the Arts student created the tweenbots which are capable of navigating long distances in urban settings relying heavily on the kindness of strangers to point them in the right direction.

If you thought that the robots would never reach their destination, think again. With the help of other pedestrians, tweenbots have successfully navigated considerable distances such as crossing the Washington Square Park in 42 minutes.

The video below shows the tweenbots in action crossing the park. Go little guy, go!

Humanoid robots to explore the Moon

2009-04-05T01:19:00.000-07:00

Japan announced the other day that they are creating a roadmap for sending humanoid robots to the Moon a a first step towards human-robot space exploration. The current roadmap as announced by Japan's Strategic Headquarters for Space Development includes sending the first humanoids to explore the Earth's natural satellite by 2020. This is only 11 years from now which makes Japan's project a very bold one. Humanoid robots can hardly navigate man-made environments with lots of structure on Earth, it is hard to believe that in the next decade robotics technology will advance so far that such robot astronauts will be possible.

I am not saying that bipedal robots will never be constructed and possibly become a major component of space missions. What I am saying is that I don't think this is going to happen by 2020. NASA has had a space robotics program for ever and the most advanced and reliable robots that they have been able to send to other planets, namely Mars, have all been wheeled rovers.

The next generation planetary explorers that NASA is preparing are also of the same kind albeit much larger and more mobile than the current and past generations. The ATHLETE rover for example is under heavy development as a future platform for exploring the Lunar surface and it's design makes much more sense than a bipedal machine. Also, don't forget about Google's Lunar X Prize which is an effort to encourage private industry to construct and send a small exploration rover to the Moon as cheaply and as quickly as possible.

At the end of the day, a rover has higher mobility, can carry a larger payload (instruments, batteries, solar panels) and can reliably navigate terrain of all kinds. To expect a bipedal robot to have such capabilities in a decade requires much ambition and lots of engineering breakthroughs.

I have a feeling that Japan is aiming for publicity more than practicality in space robotics. They have now scrapped plans for their manned space program and they have yet to start the construction of their Lunar lander. They believe it will take another 2 years before they figure out all the details regarding the announced robotics mission to the Moon. Maybe they are trying to get the media to talk about their robotics project and ignore that at the same time they announced plans for the construction and deployment of military satellites. Finally, Japan is also planning to create robotics technologies to help cleanup some of the debris floating in orbit around the Earth. This space garbage poses a huge hazard to satellites, the International Space Station, and any other man-made object that is sent into space.

Rescue robot that can clear debris and lift heavy weights

2009-03-29T01:41:00.000-07:00

Rescue robots are one application for which there has been much excitement during the last decade. These robots are designed to be small and versatile carrying a comprehensive sensor payload in order to detect victims under heavy debris in disaster areas. Everyone who has watched the news after a major earthquake or hurricane with many buildings destroyed can easily understand the need for such robots as rescuers are frantically searching for survivors under heavy debris.

Researchers from the Tokyo Institute of Technology have recently proposed a new type of rescue robot that is capable of not only detecting victims in need of help but also clearing and lifting heavy debris to reach them.

The prototype robot named Bari-bari-II has a unique design that allows it to navigate over and lift debris. Its front is designed to have a step structure which can grip on debris, lift it and move under it. Once under, the robot uses oil hydraulic power to lift up to 600Kgrs. Like traditional rescue robots, a sensor payload consisting of a camera and microphone help rescuers to find victims in the rumble. The robot weighs 25Kgrs and it measures 48x28x14cm in size. Rescuers can use more than one robots at the same time to lift even heavier debris.

The video below gives an overview of Bari-bari-II rescue robot showing it in action in a simulated disaster situation.

Game AI: The trade-off between entertainment and believable AI

2009-03-22T00:06:00.000-07:00

I don't write much about game AI since there is hardly anything cutting edge in game AI to talk about. More than a year ago, I wrote a post commenting on some research out of the Austrian Research Institute for Artificial Intelligence showing that even irrational players could easily defeat the AI in the real-time strategy game Age of Mythology. I have been playing video games for almost 20 years now and every year people talk about how the new games will have so much better AI and every year game developers fail to deliver. The truth is, that the kind of AI that goes into a computer game has different requirements than the AI systems researchers are trying to build for solving real-world problems. The differences is what makes creating good game AI systems difficult.

Game AI has to be good enough to make the game challenging for the person who paid for it but also less than optimal in its decision making so that it is not unbeatable. Nobody would enjoy playing chess against Deep Blue that even Kasparov could not defeat; it just would not be fun for anyone and let's face it, computer games are designed to entertain more than anything else. So, game developers have to create AI systems that fulfill the above requirements something that is not an easy task.

Gamasutra recently republished an article that appeared in Game Developer magazine on the subject of creating AI systems that make mistakes giving the human opponent a chance to win the game but at the same time without making the game AI look totally stupid. Mick West, the article's author, explains how one way to make AI systems to be both challenging and entertaining is by simply limiting the computational resources available to them during gameplay.

The simplest way to introduce stupidity into AI is to reduce the amount of computation that it's allowed to perform. Chess AI generally performs billions of calculations when deciding what move to make.

The more calculations that are made (and the more time taken), then (generally) the better the computer will play. If you reduce the amount of calculations performed, the computer will be a worse player.

He argues that the above method which is commonly used in games is not easy to work with because developers have a hard time finding the correct threshold for computation time. The alternative method that Mick proposes is one that has the game AI intentionally playing poorly at times to give the human player a chance at winning.

The computer has to throw the game in order to make it fun. When you beat the computer, it's an illusion. The computer let you win. We just want it to let you win in a way that feels good.

AI programmers need to get used to this idea. We are manipulating the game, creating artificial stupidity, fake stupidity. But we are not predetermining the outcome of the game.

We don't set our AI with the intent to lose the game, but rather to give the human player a reasonable chance of winning. If the human plays poorly, the AI will still win, but the player will at least feel like she came close to beating a strong opponent, and thus feel like playing one more game.

The AI engine of the chess game Fritz was designed in such a way that the game will try to create situations that a smart chess player could exploit to earn an advantage over the computer and as such have a fair chance of winning. According to Mick, the game AI is not limited in the available computational resources but instead requires more in order to do this extra bit of thinking such that the game is both entertaining and challenging.

The article which you can read here, gives more examples of game AI employing the above tactic for poker and snooker.

Lockheed Martin exoskeleton: The Human Universal Load Carrier (HULC)

2009-03-14T01:33:00.000-07:00

In the span of less than a year, exoskeletons have moved from science fiction to reality. In the last few months, several research and developments groups have unveiled working prototypes of exoskeletons. One example is the Sarcos exoskeleton designed for military use; another is Honda's robotic legs recently announced; third, Cyberdyne's HAL exoskeleton from Japan is available commercially for citizens with medical needs at a price of a few thousand dollars a month. Recently, Lockheed Martin unveiled their own exoskeleton for the military. The Human Universal Load Carrier (HULC) is under development for military use.

This is how Lockheed describes the titanium-made HULC,

The HULC is a completely un-tethered, hydraulic-powered anthropomorphic exoskeleton that provides users with the ability to carry loads of up to 200 lbs for extended periods of time and over all terrains. Its flexible design allows for deep squats, crawls and upper-body lifting. There is no joystick or other control mechanism. The exoskeleton senses what users want to do and where they want to go. It augments their ability, strength and endurance.

Soldiers wearing the exoskeleton can reach running speeds of up to 10 miles per hour and it only takes about 30 seconds to put it on or take it off. The company also claims that the HULC can assist in heavy lifting even when it runs out of power; I don't know how that works but it sounds impressive. The exoskeleton can be enhanced with a number of accessories that permit soldiers to easily lift heavy weights including a swat ballistic shield.

The promotional video below showcases the exoskeleton and all its features. Future soldiers will certainly look very different compared to today going to war equipped with all kinds of state-of-the-art technology to assist them. I just hope that such great technologies will not only be used to kill others but also (or better yet only) help those who need them the most.

Wolfram Alpha computational knowledge engine

2009-03-10T01:44:00.000-07:00

Stephen Wolfram announced a couple of days ago in a blog post that he and his team will soon unveil a new online computational knowledge engine called
Wolfram Alpha that will be able to answer questions posed to it in natural language. If successful, this would be the holly grail of computational intelligence for no other reason that questions posed in natural language could be directly answered by the system as opposed to simply returning online documents that might include the answer; the latter is the way that Internet search engines such as Google work.

Granted, for some questions, Google is very capable of directly giving answers such as for example asking “what is the population of China?” But whereas Google's answers are likely found in some online document that directly answers the query, Wolfram Alpha will compute the answer by deriving it from its computational knowledge base. In other words, Wolfram Alpha will be able to derive new facts from known facts and a set of rules. This is what Artificial Intelligence researchers have dreamed of being able to do for more than 50 years.

Many details of how Wolfram Alpha is going to work were not provided in Stephen Wolfram's blog post. In just a few sentences, this is how he describes the inner workings of the computational knowledge engine,

So how can we deal with that? Well, some people have thought the way forward must be to somehow automatically understand the natural language that exists on the web. Perhaps getting the web semantically tagged to make that easier.

But armed with Mathematica and NKS (New Kind of Science) I realized there’s another way: explicitly implement methods and models, as algorithms, and explicitly curate all data so that it is immediately computable.

It’s not easy to do this. Every different kind of method and model—and data—has its own special features and character. But with a mixture of Mathematica and NKS automation, and a lot of human experts, I’m happy to say that we’ve gotten a very long way.

I wasn’t at all sure it was going to work. But I’m happy to say that with a mixture of many clever algorithms and heuristics, lots of linguistic discovery and linguistic curation, and what probably amount to some serious theoretical breakthroughs, we’re actually managing to make it work.

You can read the Wolfram's entire post here.

Wolfram Alpha will go online in May of 2009, just 2 months from now. I am looking forward to it. I don't expect it to be perfect but I trust those people with access to the alpha version who say that it works very well. The only thing I have to figure out over the next two months is what to ask it when I finally get my chance. Maybe the answer to my question will be 42.

MITRE immersive spherical vision system

2009-03-01T00:59:00.000-08:00

Immersive vision systems for teleoperation are a valuable tool for many applications including inspections of old buildings, pipelines and sewages, search and rescue, and military, i.e., detection and neutralization of roadside bombs. Immersive systems work by presenting a virtual representation of the world seen via a camera that is situated away from the operator (note: I use the term virtual representation of the world liberally here because what the operator sees is actually images of the real world captured by a remote camera; I call it virtual because a explained later these images arrive delayed which means that the real world may have change since the data was collected.)

The operator views slices of the virtual world using a head mounted display while a sensor detects his movements and updates the view accordingly. A large problem with such methods is the latency between the operator moving his head and the system updating his view; the latency often comes from the fact that the remote camera has a limited field of view meaning that every time the operator moves his head, a pan-tilt mechanical unit has to reposition the camera delaying the relay of the images and making it difficult to operate the remote system often causing lots of distraught for the operator.

MITRE scientists have worked out a solution to this latency problem replacing the limited field of view camera on the pan-tilt unit with a spherical vision camera which has no moving parts. The camera of choice is the Ladybug commercially sold by Point Grey Research in Canada. Stanford's Urban Challenge autonomous car also used the same camera for part of its perception system. This spherical vision system consists of 6 cameras which capture images simultaneously covering a large portion of the view sphere around it. Software stitches the images together into a single view in real-time. These spherical vision images are then available to the operator to view for any orientation of his head. The latency I mentioned earlier is thus eliminated by the fact that the camera need not be repositioned every time the operator moves his head. Moreover, more than one operators can be using the system looking in different directions.

The below promotional video shows the capabilities of the immersive spherical vision system including some of its potential applications. The true power of the system is clearly visible in the part of the video where a car driver is shown driving down a street while perceiving the world in real-time via a head mounted display.