Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
9654548 | Robotics and Autonomous Systems | 2005 | 19 Pages |
Abstract
In this paper, we propose a control strategy allowing us to perform the dynamic walking gait of an under-actuated robot even if this one is subjected to destabilizing external disturbances. This control strategy is based on two stages. The first one consists of using a set of pragmatic rules in order to generate a succession of passive and active phases allowing us to perform a dynamic walking gait of the robot. The joint trajectories of this reference gait are learned by using neural networks. In the second stage, we use these neural networks to generate the learned trajectories during the first stage. The goal of the use of these neural networks is to increase the robustness of the control of the dynamic walking gait of this robot in the case of external disturbances. The first experimental results are also presented.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Christophe Sabourin, Olivier Bruneau,