Model-free distributed coordination control for dual-arm reconfigurable manipulators with handling task constraints

In this paper, a distributed coordination control of dual-arm reconfigurable manipulators based on neural networks learning is presented for handling tasks. Based on Newton-Euler algorithm, the dynamic models of the manipulators and grasped object are established through kinematic analyze, respectively. According to the load distribution method, the motion-induced force is effectively distributed to each arm of the manipulator, and then the dynamics of the single reconfigurable manipulator is obtained. An improved sliding mode function is designed to reflect the error of the joint trajectory and internal force. The radial basis function neural network (RBFNN) is utilized to learn the uncertain dynamics of reconfigurable manipulator. Then the model-free distributed coordination controller is obtained. The asymptotic stability of dual-arm reconfigurable manipulators is proved through Lyapunov stability theory. Finally, the validity of the model-free distributed coordination controller is verified by simulations.