A primal-dual neural network for joint torque optimization of redundant manipulators subject to torque limit constraints

A primal-dual neural network is proposed for the joint torque optimization of redundant manipulators subject to torque limit constraints. The neural network generates the minimum driving joint torques which never exceed the hardware limits and make the end-effector to track a desired trajectory. The consideration of physical limits prevents the manipulator from torque saturation and hence ensures good tracking accuracy. The neural network is proven to be globally convergent to the optimal solution. The simulation results show that the neural network is capable of effectively computing the optimal redundancy resolution.