Abstract: It is significant to improve flexibility of the manipulator and avoid singularity in the motion planning and control. An effective method is to maximize the manipulability index. However, the manipulability index is a non-convex function of joint angle, and optimizing this index is a great challenge. Therefore, an acceleration level manipulability maximization (ALMM) method is proposed. Firstly, the nonlinear optimization problem in position level is transformed to velocity level and acceleration level by a new multi-level simultaneous minimization scheme, and the problem of maximization manipulability is reconstructed in acceleration level. Secondly, the minimum velocity index is introduced to ensure the stability of the system and keep the joint angular velocity at a low level. Then, combined with the physical limits of the joint, the above problem is formulated as a standard quadratic programming (QP) problem to solve. Taking SSRMS configuration redundant space manipulator as an example, the simulation results show that the ALMM method can achieve the optimization of manipulability while completing the trajectory tracking task, and the comparative experiment proves the effectiveness and superiority of the ALMM method.

Key words: redundant manipulator, inverse kinematics, manipulability, quadratic programming (QP)