A fully neural-network-based planning scheme for torque minimization of redundant manipulators

The aim of this paper is to develop a new method for minimizing joint torques of redundant manipulators in the Chebyshev sense and to present a fully neural-network-based computational scheme for its implementation. Minimax techniques are used to determine the null space acceleration vector which can guarantee to minimize the maximum joint torque. For real-time implementation, we transform the proposed method into a computation of a recurrent neural network. At each time step, the neural network is adopted for both the solution of the least-norm joint acceleration and the determination of the optimum null space acceleration vector. Compared with previous torque minimization schemes, the proposed method enables more direct monitoring and control of the magnitudes of the individual joint torques than does the minimization of the sum of squares of the components. Simulation results demonstrate that the proposed method is effective for the torque minimization control of redundant manipulators.