Bipedal-double-pendulum walking robot control using recurrent hybrid neural network

Abstract

This chapter presents neural control scheme ofa planar-like double-pendulum-bipedal robot. For simplicity, only a five-link planar system is considered. The system effectively acts as two dynamically interacting planar robot arms. The scheme employs a single neural controller for the whole biped. Recurrent networks have feedback connections and thus an inherent memory for dynamics which makes them suitable for dynamic system modeling. A feature of the networks adopted is their hybrid hidden layer which includes both linear and nonlinear neurons. The standard proportional derivative (PD) controller is also employed for comparison. The results presented show the superior ability of the proposed neural control scheme at adapting to changes in the dynamics parameters of the bipedal robot.