Decentralized Guaranteed Cost Control for Reconfigurable Manipulators with Environmental Constraints Based on Adaptive Dynamic Programming

In this paper, a decentralized guaranteed cost control method based on adaptive dynamic programming (ADP) is proposed to address the optimal control problem of reconfigurable manipulators with environmental constraints. The designed control policy is adopted to deal with the effect of joint friction, interconnection dynamic coupling (IDC) and uncertain external environmental constraints. Based on the ADP algorithm, an online policy iteration algorithm is used to solve the Hamilton-Jacobi-Bellman (HJB) equation with a modified cost function, and the approximate optimal control strategy is obtained by constructing a critic neural network that is used to approximate optimal cost function. The closed-loop robotic system is proved to be asymptotic stability by using the Lyapunov theory. Finally, the simulations of the manipulators with two different configurations are given to illustrate the effectiveness of the proposed method.