Bi-criteria torque minimization of redundant robots performing different end-effector trajectories by using simplified LVI-PDNN

In this paper, we present a bi-criteria torque-minimizing (BCTM) scheme for robotic redundancy resolution. Such a bi-criteria scheme combines the minimum two-norm and minimum infinity-norm solutions of joint torque via a weighting factor. In addition, the presented BCTM scheme is unified as a general quadratic program (QP) subject to equality, inequality, and bound constraints. As a real-time QP solver, a simplified linear variational inequality (LVI) based primal-dual neural network (PDNN), i.e., a simplified LVI-PDNN, is then exploited to solve online the BCTM scheme. Computer-simulation results based on PUMA560 robot manipulator tracking different types of end-effector trajectories further verify the effectiveness and flexibility of the presented BCTM scheme.