| Presentation | 2005/7/20 Learning of Qusasi-Passive Dynamic Walking by a Stochastic Policy Gradient Method Kentarou HITOMI, Tomohiro SHIBATA, Yutaka NAKAMURA, Shin ISHII, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | A class of biped locomotion called Passive Dynamic Walking (PDW) has been recognized to be efficient in energy consumption and a key to understand human walking. Although PDW is sensitive to the initial condition and disturbances, some studies of Quasi-PDW, which incorporates supplemental actuators, have been reported to overcome the sensitivity. In this article, we propose a reinforcement learning scheme designed in particular for Quasi-PDW walking. The keys of our approach are a reward function and a learning method of a simple intermittent feedback controller, both of which utilize the robot's passive dynamics as much as possible. They successfully make the action selection problem for walking significantly reduced. Computer simulations show that the parameter in a Quasi-PDW controller is quickly learned after only 180 episodes, and that the obtained controller is robust against sudden perturbations and variations in the slope gradient. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | passive dynamic walking / reinforcement learning / adaptive control |
| Paper # | NC2005-35 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2005/7/20(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Neurocomputing (NC) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Learning of Qusasi-Passive Dynamic Walking by a Stochastic Policy Gradient Method |
| Sub Title (in English) | |
| Keyword(1) | passive dynamic walking |
| Keyword(2) | reinforcement learning |
| Keyword(3) | adaptive control |
| 1st Author's Name | Kentarou HITOMI |
| 1st Author's Affiliation | Graduate School of Information Science, Nara Institute of Science and Technology() |
| 2nd Author's Name | Tomohiro SHIBATA |
| 2nd Author's Affiliation | Graduate School of Information Science, Nara Institute of Science and Technology:Computational Neuroscience Laboratories, ATR |
| 3rd Author's Name | Yutaka NAKAMURA |
| 3rd Author's Affiliation | Graduate School of Information Science, Nara Institute of Science and Technology |
| 4th Author's Name | Shin ISHII |
| 4th Author's Affiliation | Graduate School of Information Science, Nara Institute of Science and Technology |
| Date | 2005/7/20 |
| Paper # | NC2005-35 |
| Volume (vol) | vol.105 |
| Number (no) | 211 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |