If you've been impressed with Honda's ASIMO robot, get ready for something far more awesome.
Meet the first prototype of PETMAN, the new bipedal robot being developed by Boston Dynamics. PETMAN, as we discussed previously, is designed to test chemical warfare suits by imitating the same range and speed of motion as a human being -- walking, running, climbing, crawling, and so forth.
What you're seeing here is the first released video of PETMAN's proof-of-concept prototype. Using much of the same hardware as the famous BigDog robot, they've developed a self-balancing (but currently externally powered) prototype that walks on a large treadmill. The prototype has a top speed of 3.2mph -- well over Asimo's top walking speed, and nearly its running speed. It also appears that the walking speed adapts automatically to the speed of the treadmill.
You'll also note that those stylin' climbing shoes highlight a heel-toe walking pattern, a big dynamic change from the sort of "hoof" that BigDog has.
I can't wait to see how the actual robot performs. This prototype is already impressive.
We were invited to Stanford on Thursday for a sneak peak at their latest robot car, from the family that includes Stanley and Junior. It’s an Audi TTS that’s been modified with sensors, GPS guidance, and a trunkfull of computers, but it’s not intended to drive you to work in the morning… It’s actually a race car, designed to push the limits of driving performance. Already, this TTS holds the unofficial world speed record for an autonomous car at 130 kph (edit- they meant to say 130 mph, which is a lot more impressive), but it’s capable of a whole lot more. Basically, Stanford is figuring out how close to the edge of control a car can be driven, and then they’re going to program their Audi to drive on that edge. They’ve set themselves a challenge of racing to the top of Pikes Peak sometime next year:
Well, let's say the video above shows the robot's first runs on the slopes. This mechanical skier has been practicing, and it now can race downhill and even make turns to pass between gates, slalom-style.
That's what Bojan Nemec from the Jozef Stefan Institute in Slovenia told a packed house at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) on Tuesday.
Nemec said his goal is to design a robot skier capable of autonomous skiing using the carving technique. (Apparently a robot can't ski using regular technique; it's too hard. But with carving, the skis practically ski themselves, according to Nemec.)
This isn't the first skiing robot, but it's bigger and heavier than earlierJapanesemodels, Nemec explained in his talk. Ideally, a skiing robot would be able to use off-the-shelf skis, rather than custom-made miniature ones.
His robot can. About the size of an eight-year-old child, the skiing bot looks a lot like a laptop on legs. Nemec got its skis at the local ski shop.
Here's a video describing the project:
The laptop control center plans the robot's trajectory, using a camera to measure its distance from the race gates. Gyros and force sensors help the bot stay stabilized on the slopes.
The robot carries a GPS unit, but it's used to help measure speed, not for trajectory planning. That makes sense, if you're trying to build a robot that works more like a human, relying on vision.
Sometimes their robot got away from them:
And sometimes, as Nemec said… "Well, it happens to the best!"
Here’s a sample of questions from the audience, which Nemec answered with great aplomb:
Can it stop? "Yes, of course," Nemec said as he paused the video that was running. The bot stopped. Everyone laughed. But yes, it can stop, sort of.
What would it need to be able to compete with humans in a downhill skiing race in, say, 2015? Pause. "I think people are expecting too much from this robot." Laughs. "But it would need additional degrees of freedom, and should be more robust -- once our robot escaped from our control, and it broke a lot of parts. So, more robust."
Why do we need a skiing robot anwyay? "Testing of ski equipment, or modeling of skiing for VR applications." Hmm. What about using it as a robot teacher, i.e. ski instructor, or for inspecting the ski slopes? Nemec was skeptical about those applications: "Not in the near future."
Halloween is approaching, so how about ... DIY cockroach robot!
The Dynamic Autonomous Sprawled Hexapod, aptly abbreviated DASH, really moves. It's a high-speed six-legged runner that can be built in an hour using basically cardboard and polymer sheets for its frame.
Well, it helps if you have a laser cutter and a PhD in robotics.
Created by Paul Birkmeyer and Prof. Ronald Fearing at the Biomimetic Millisystems Lab at UC Berkeley, DASH is extremely lightweight (16 grams) and uses a single DC motor to power the legs and a small servomotor to slightly deform the robot's body, making it turn left or right.
The little robot can reach speeds of 1.5 meters per second and is flexible/strong enough to be dropped from a height of 28 meters without breaking. It picks up and dashes off again.
Just be careful about running the robot near people who are squeamish about insects -- or DASH might get smashed.
This is by far one of the coolest and weirdest robot prototypes we at IEEE Spectrum have ever seen.
Meet iRobot's soft, shape-shifting robot blob. It rolls around and changes shape, and it will be able to squeeze through tiny cracks in a wall when the project is finished.
(Skip the first 1:50 minutes of the video above to see the blob in action.)
Researchers from iRobot and the University of Chicago discussed their palm-sized soft robot, known as a chemical robot, or chembot, at the IEEE/RSJ International Conference on Intelligent Robots and Systems yesterday. It's "the first demonstration of a completely soft, mobile robot using jamming as an enabling technology," they write in a paper presented at the conference.
The concept of "jamming skin enabled locomotion" is explained quite nicely in the video. The polymer used for the bot’s stretchy skin is off-the-shelf silicone two-part rubber.
By controlling the parts of the blob that "inflate," the researchers can make it roll.
The video shows the project as it was about a year ago. The current stage has a bit different design and is moving toward the ability to include sensors or even connect different blobs together, but those details are sketchy.
When asked about the usefulness of such a bot, iRobot researcher Annan Mozeika promptly answered, "to squeeze into small holes." And who wants to do that? DARPA, of course. End of questions.
Update: We corrected a typo above: silicon should be silicone -- hah!
When I think about robots, I don't automatically think of cockroaches. But the little creepy crawlies were featured in several presentations at yesterday's workshop on Biologically Inspired Robotics at the 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems, which officially begins today in St. Louis, Missouri.
The highlight (at least for me) was a debate between Case Western engineering professor Roger Quinn and Singapore's Nanyang Technological University professor K.H. Low about which cockroach was faster and smarter: the North American cockroach, or the Asian cockroach.
Low claimed the Asian cockroach was faster, since he could never catch one at home, while he caught them all the time during six years of study in Canada. Quinn begged to differ, and the ensuing discussion of roach biology entertained the laptop-bedecked crowd of about 50 for several minutes.
Unfortunately no live demonstrations could be had, but Quinn seemed excited about the possibility of a race.
Biologically inspired robotics is a hot topic at this conference, with one of 16 simultaneous tracks devoted to it, and several more touching on its themes. Other tracks will hit human-robot interaction, medical robots, legged robots, and underwater robots, to name just a few.
Sangbae Kim, an assistant professor at MIT and director of MIT’s Biomimetic Robotics Lab, is looking forward moving beyond cockroach-inspired design, and even past the gecko-inspired bots he worked on while at Stanford.
Kim's research at MIT is now focusing on "hyper dynamic locomotion." Translation? Cheetah-bot.
"You see so many robots everywhere," Kim says, "especially at conferences like these. But none of the robots can follow me, at walking speed, over rough terrain, or up stairs."
Kim thinks that kind of dynamic locomotion is "really limited still," and the robots that can crawl over rough terrain are small and often go slow. So he's aiming for a bigger, legged robot that could run fast, like a cheetah.
"Not necessarily over rough terrain," he says, "but over flat and also rough" spaces. And not necessarily even a cheetah, he says, not wanting to limit his goals. Squirrels, too, have highly dynamic behavior that allows them both to run fast and to climb, actions that can only be accomplished by two very different kinds of bots so far.
It's a good reminder that robots still can't do nearly the variety of tasks that humans and animals can. But presenters and exhibitors at this conference will set out to prove how much their robots can do.
Above is one of the videos presented today featuring a cockroach-like critter bot.
Imagine purchasing a new Rovio robot. This wheeled mobile robot sports a webcam and can be accessed easily through the internet. Often these and other robots are bought as toys, used by the owners to check on their home during a vacation, perhaps for teleconferencing, or to check on an elderly loved one.
Now imagine a malevolent hacker from Russia or China, or your next door neighbor, or a even stalker gaining access to this robot. Now they have free access to your home, roving about checking to see if the owner is home, spying on your children, or perhaps taking embarrassing video of you or your family. What if the robot is commanded to break items in your home, hide your keys, or drive under the feet of granny to harm her? Millions of these robots have been sold, meaning they are quite ubiquitous and therefore prime targets of malicious hackers.
Researchers at the University of Washington recently studied the RoboSapien V2, the Rovio, and the Spykee and found quite a few easily exploitable security flaws. Although today's robots are relatively harmless and limited, it points out how security features are generally lacking in the design. Understandable, considering how cost is very much a key factor for a robot succeeding in the market place. Yet robots in the future will be stronger, more capable, and unfortunately perhaps even more exploitable.
Security was an afterthought for the design of the internet. It doesn't need to be for the coming robot revolution.
You can find a quick summary of the findings here:
The robots developed by Thomas Bewley and his team at the Coordinated Robotics Lab at the University of Califonia, San Diego, may look rather simple at first. But it turns out these 'bots are capable of impressive acrobatic maneuvers.
The Switchblade rover can balance on the tip of its treads and climb stairs by flipping itself end-over-end. iHop balances itself by using its wheels as gyros and it can hop on its pogo-stick of a leg.
Check out the video below by Spectrum's Josh Romero, showing how the UCSD engineers are giving their robots new ways to move.
One of the innovations featured at this year's CEATEC exhibition, running from October 6th to 10th at the Makuhari Messe just outside Tokyo, is a new robot car concept by Nissan. At 50cm in height and with a maximum speed of approximately 1.4km/h the EPORO robots can hardly be called car demonstrators, but the small robots moving in a group of 6 showcase some interesting technology.
Equipped with laser range finders and UWB communication the robots use a swarming algorithm similar to that observed in fish. While this is hardly new - computer simulations of such swarming behaviors have been conducted since the 80s - what's interesting is the implementation of a real demonstrator by a leading car manufacturer.
Like real fish, the EPORO only relies on relative positioning which would allow the necessary scalability for such a technology to be implemented in real cars.
While Nissan has not disclosed the exact algorithm, it is based on a basic shoaling algorithm, splitting the behaviour into three modules based on distance as illustrated in the right figure above:
- AREA 1: Collision Avoidance: Change traveling direction without colliding with other fish.
- AREA 2: Traveling Side-by-Side: Travel side-by-side with other fish while keeping a certain distance between each fish (to match the speed).
- AREA 3: Approaching: Gain closer proximity to other fish that are at a distance from them.
Nissan does not necessarily plan to commercialize the little robots themselves but aims at employing the underlying technology in their future vehicles by integrating it into their Safety Shield concept, aimed at supporting drivers in all situations from normal driving to the events following a collision.
Following an earlier prototype which could just avoid collisions last year and this year's announcement of the EPORO, it will be interesting to see Nissan's next step in moving these technologies towards commercialization.
Several months ago we mentioned that Boston Dynamics had received a grant to work on a new version of the "Precision Urban Hopper", a small wheeled robot designed to hop over obstacles 40-60 times its size. Working with Sandia National Labs, they've created a demonstration platform using the hopping mechanism whose demo has been making its way around YouTube. But: why is this demonstration important?
A lot of the coverage I've seen has mentioned that this could be a "PackBot killer" -- suggesting that it may compete with iRobot's highly successful millitary platform, or the similar Talon robot from QinetiQ. Though the platforms have a common shape, I don't think this is the interesting thing about this. The platform is designed specifically to demonstrate the hopping mechanism, and it carries no other payload -- no teleoperated arm to disarm IEDs, no weapons, none of the sensor payloads found on the iRobot and QinetiQ packages. What I think we'll see instead is the development of this mechanism for installation on platforms like PackBot -- or, more likely, SUGV -- and a similar version of the Talon. iRobot has always had videos showing PackBots that can be thrown through a window and be able to immedately start rolling around in a building. It seems like a natural extension of this is a SUGV that can hop up through a second-story window, right itself, and perform its mission.