Critic Only Policy Iteration-based Zero-sum Neuro-optimal Control of Modular and Reconfigurable Robots with uncertain disturbance via Adaptive Dynamic Programming

Abstract

A critic only policy iteration (COPI) scheme-based zero-sum neuro-optimal control method has been presented via adaptive dynamic programming (ADP) to address optimal trajectory velocity and tracking control of modular and reconfigurable robots (MRRs) problem. Based on policy iteration (PI) and ADP method, Hamilton-Jacobi-Issacs (HJI) equation is addressed by using only critic neural network (NN). The approximated optimal control can be obtained. Closed-loop system is proved to be asymptotic stable according to the Lyapunov theory. At last, simulations are demonstrated to show effectiveness of method.