The Best Robots of 2008 57
An anonymous reader writes "Robot innovation continued its relentless advances during 2008. SingularityHub has a showcase of the best robot videos of the past year. These robot videos are really amazing, and they show just how far we have come in the field of robotics in recent years."
The videos include toy robots, robot musicians (which we've discussed in the past), modular robots that work together to move around, robots doing synchronized martial arts, the BigDog robot that can walk on almost any type of terrain, and robot soccer. We've also recently talked about a couple of robots that will bring you beer.
Error establishing a database connection (Score:2, Informative)
Can't connect to database? (Score:2, Informative)
The server seems to be running on a robot... and the robot has left the building. Impressive.
Re:There's been real progress (Score:3, Informative)
No offense to DARPA, not all of the navigation and vision algorithms in those cars with a whole set of high speed computers are really practical for use on smaller home service robots.
Vision works better on home service robots that it does outdoors. Outdoors, getting a long enough baseline for a stereo pair is hard, except through motion vision. Humans only have stereo out to a few meters, anyway. SLAM (Simultaneous Location and Mapping) for mobile robots is getting quite good. Willow Robotics demoed their system at RoboDevelopment a few months ago, and the latest issue of IEEE Trans. on Robotics, a special issue on SLAM, indicates how good that's become.
But machine learning is facing some strong limitations when compared with the abilities of biological systems in coping with unsupervised learning in uncertain and dynamic environments.
I recently went over to Stanford to see the CS229 project presentations [stanford.edu], and it's very impressive what small teams of students are getting done in one quarter. Self-guiding robot helicopters, for example. The field has moved away from neural nets; Bayesian statistics, with real theory underneath, works better.
the balance and slip control of Big Dog, applies to quadrupeds with the similar mechanical characteristics. If you are trying to imply that the results are relevant to humanoids, I suggest you read up on the loads of material on everything from 3d linear inverted pendulum model to spin angular momentum regulation and control for humanoids.
Been there, done that, own the patent on legged slip control. [animats.com] For systems which really use dynamic balance, the number of legs doesn't matter all that much from an algorithm standpoint. In fact, most real progress has been made by first getting the one-legged case to work. Key insights: 1) balance has priority over movement, 2) slip/traction control has priority over balance, 3) legs need three joints, not two, so you can play with the force vector at ground contact independent of foot position, and 4) legs are viewed as assets to be deployed to manage traction, balance, and propulsion. "Gaits" are an emergent behavior, the state into which things settle down when movement is not disrupted.