The Massachusetts Institute of Technology
|
The Massachusetts Institute of Technology
|
|
To model the neural networks of the brain using mathematical theories,
computer simulation, and circuits of biological neurons in vitro.
[Home]
[People]
[Projects]
[Courses]
[Contact]
[Hiring]
[Media]
[Links]
[Pictures]
Lab-members only [html|wiki]
 |
Russ Tedrake Graduate Student |
Reinforcement Learning for Dynamically Stable Legged LocomotionFor a bipedal robot to walk like a human, it must constantly move itself through positions that defy conventional theories of robotic stability. Furthermore, this balance has to be maintained while iteracting with an uncertain ground and possibly colliding with unexpected obstacles. Yet somehow humans solve the same problem almost effortlessly. Presumably this is because of both a superior mechanical design and a superior, adaptive control strategy.In my research, I am using reinforcement learning and ideas from biological motor control to create control algorithms for the walking robots in the MIT Leg Laboratory.
Computational Motor ControlI am generally interested in motor control problems where biological systems outperform their robotic counterparts. Please check out the Computational Motor Control Journal Club for more information. |
|
Last Updated on Thursday, 08-Aug-2002 10:09:48 EDT