On Maximizing Manipulability Index while Solving a Kinematics Task

In this paper, we investigate the problem of maximizing the manipulability index while solving a general Inverse Kinematics (IK) problem of a redundant industrial manipulator. Manipulability index has been extensively studied in the robotics literature and several formulae have been developed, nevertheless, they mainly only exploit the robot redundancy. The general IK is formulated as a Quadratic Programming (QP) that can seamlessly incorporate inequality constraints, such as collision avoidance, and we propose two new formulae to integrate the manipulability index maximization into the QP-based IK solver. We then thoroughly analyze the performance of the proposed formulae in simulation and validate them on a real Baxter research robot. The experimental results revealed the outperformance of the proposed formulae in comparison with the classical formula in the literature. Hence, providing a way to improve the manipulability index of a recorded trajectory, e.g. by learning by demonstration, or an offline generated one by a motion planning algorithm.