A simulation study of bipedal walking robots: modeling, walking algorithms, and neural network control

Abstract

The purpose of this thesis is to develop walking algorithms for use with mechanical bipeds. This thesis is comprised of three parts. A two dimensional biped simulator, called WALK, is developed. This simulator is designed to facilitate the evaluation of bipedal walking algorithms. Then, a two dimensional walking algorithm is developed using a simple inverted pendulum model. This algorithm is shown to provide for stable walking using the WALK simulator system. A neurocomputer adaptive controller that is based on the CMAC architecture is added to the inverted pendulum model. The adaptive walking algorithm is not only stable, but provides for accurate control when leg inertia and modeling errors are considered.