Adaptive control of biped robot walking on an inclined plane

Abstract

The problem of walk control of a biped robot is considered. The inertial robot parameters are assumed unknown and control actions are applied only twice during one step. The control aim is for robot walking with a given average speed without falls and jumps; a desired trajectory is not given. A two-level control method is designed and theoretically justified. The lower level is an adaptation block based on the recurrent aim inequality method. The higher level is a "trajectory calculating" block using the original global optimization method. The results of computer simulation confirm the necessity of adaptive control and efficiency of the algorithms.