| Presentation | 2007-03-14 Learning of a robust controller for a biped robot based on a sample-reuse reinforcement learning method Tsuyoshi UENO, Yutaka NAKAMURA, Takashi TAKUMA, Tomohiro SHIBATA, Koh Hosoda, Shin ISHI, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Recently, many researchers on humanoid robotics are interested in Quasi-Passive Dynamic Walking (Quasi-PDW), which is similar to human walking. It is desirable that control parameters in Quasi-PDW are automatically adjusted because robots often suffer from changes in their physical parameters and the surrounding environment. Reinforcement learning (RL) can be a key technology to this adaptability, and it has been shown that RL realizes Quasi-PDW in simulation studies. To apply the existing RL method to controlling real robots, however, further improvement to accelerate its learning is required. Otherwise, the robots would break down before acquiring appropriate controller. For this purpose, this study employs an off-policy natural actor-critic (off-NAC) which is able to reuse the samples that has already been obtained. This study also proposes an adaptive method of the learning rate which works with the off-NAC method. Simulation as well as real experiments demonstrate that fast and stable learning of Quasi-PDW of an unstable biped robot can be realized by our proposed method. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Reinforcement Learning / Quasi-Passive Dynamic Walk / Adaptive Control |
| Paper # | NC2006-151 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2007/3/7(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 a robust controller for a biped robot based on a sample-reuse reinforcement learning method |
| Sub Title (in English) | |
| Keyword(1) | Reinforcement Learning |
| Keyword(2) | Quasi-Passive Dynamic Walk |
| Keyword(3) | Adaptive Control |
| 1st Author's Name | Tsuyoshi UENO |
| 1st Author's Affiliation | Department of Information Science, Nara Institute of Science and Technology() |
| 2nd Author's Name | Yutaka NAKAMURA |
| 2nd Author's Affiliation | Department of Engineering, Osaka University |
| 3rd Author's Name | Takashi TAKUMA |
| 3rd Author's Affiliation | Department of Engineering, Osaka University |
| 4th Author's Name | Tomohiro SHIBATA |
| 4th Author's Affiliation | Department of Information Science, Nara Institute of Science and Technology |
| 5th Author's Name | Koh Hosoda |
| 5th Author's Affiliation | Department of Engineering, Osaka University |
| 6th Author's Name | Shin ISHI |
| 6th Author's Affiliation | Department of Information Science, Nara Institute of Science and Technology |
| Date | 2007-03-14 |
| Paper # | NC2006-151 |
| Volume (vol) | vol.106 |
| Number (no) | 588 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |