Deterministic Learning Based Compass Biped Robot Control

In this paper, based on a resent result of deterministic learning, we present an approach for the compass gait biped control and learning. When a compass-like robot is controlled to perform a desired walking pattern, locally-accurate approximation of the closed-loop control system dynamics can be achieved in a local region along the periodic gait. Moreover, like the learning process of human walking, the learned knowledge can be reused for the same or similar control tasks. Compared with previous methods, the robot can be easily controlled with little effort. This paper continues our work on deterministic learning based robot manipulator control, and establishes the foundation for more complex biped robot control and learning problem. Simulation studies are included to illustrate the proposed approach.