Human-Like Motion Planning of Anthropomorphic Arms Based on Hierarchical Strategy

Abstract

The human-like motion of anthropomorphic arms can improve the efficiency of HRI (human-robot interaction). We propose a simple and easy-to-use human-like motion planning algorithm of anthropomorphic arms based on hierarchical strategy. By setting the trigger conditions of the end-effector's position and attitude, the motion of the robotic arm is divided into reaching stage and grasping stage. The end-effector velocity is planned using minimum jerk model, and a clamping function is applied to improve the smoothness of the motion. The inverse kinematics is solved using selectively damped least squares method, which can avoid singularities. To deal with the redundancy problem, a human-like pose optimization function based on the principle of minimum potential energy is defined, which is solved using gradient projection method. The simulation and experimental results show that the robotic arm achieves good human-like performance in both Cartesian space and joint space, confirming the effectiveness of the motion planning algorithm proposed in this paper.