Manipulability Optimization for Redundant Dual-Arm Robots at the Acceleration Level

Existing manipulability optimization schemes typically solve at the velocity level, which cannot consider joint acceleration limits and are unsuitable for torque control of robotic arms. Therefore, this paper constructs a cost function that considers both joint torque constraints and manipulability optimization of the manipulator, and equivalently transforms it into a convex quadratic function. The proposed scheme addresses the non-convexity issue of manipulability with respect to the robotic arm joint acceleration and the inversion problem of the generalized Jacobian matrix. Simulation results show that the proposed method can maximize the manipulability of redundant dual-arm robots at the acceleration level, verifying the effectiveness of the scheme.