Global Redundancy Optimization of Manipulability for 7-DOFs Anthropomorphic Manipulator Using Joint Monotonicity

Real-time manipulability maximization is a common sub-task for the a redundant manipulator. However, due to the heavy computational burden and non-convex characteristic of the manipulability function, most existing methods only locally maximize manipulability in redundancy under real-time conditions. To address this problem, we propose a real-time compatible method that achieves globally maximizes manipulability in redundancy, while simultaneously tracking the end-effector’s pose. Our method is particularly applicable to a 7-degrees of freedom (DOFs) anthropomorphic manipulator, where its joint exhibits monotonicity. Utilizing the joint monotonicity, we initially propose an algorithm capable of computing redundancy resolution for global maximum manipulability in real-time. Subsequently, the task of optimizing manipulability in redundancy is transformed into tracking the redundancy joint with maximum manipulability. The redundancy joint tracking task is then integrated into an optimization-based controller. The effectiveness of the proposed method is verified in simulation and experiment.