Event-triggered-based optimal control for reconfigurable robot via mixed nonzero-sum game

Abstract

This paper presents an event-triggered mixed nonzero-sum game optimal control method for reconfigurable robot (RR) under coordinated operation tasks. Firstly, the contact force on the end-effector is mapped to each joint through the joint task assignment method, and then the dynamic model of subsystem is established based on the joint torque feedback technique. For ensuring the comprehensive optimal control performance and energy consumption and realizing the optimal coordinated control scheme, two kinds of performance indexes are defined, including the nonzero-sum game among each joint module of the manipulator and the cooperative game between the end-effector and the manipulated object. Afterwards, the coordination control problem of RR system is transformed into a mixed nonzero-sum game problem among joint and manipulated object. Next, in the framework of adaptive dynamic programming (ADP) algorithm and event-triggered mechanism, a single critic NN with learning structure is utilized to solve the event-triggered version Hamiltonian equation. Finally, experimental results verify the effectiveness of the developed method.