Predictive Damped Inverse Kinematics for Redundant and Underactuated Robotic Systems

Abstract

This paper presents an effective and efficient online inverse kinematics method with feedback correction for redundant and underactuated robotic systems by means of linear model predictive control. The physical limitations of robotic systems are encoded in the constraints of the quadratic program which is calculated at each control update time. To avoid the effects of singularities, a damping item is introduced in the stage cost function of the quadratic program. Besides, a weight strategy for control allocation is encoded in the cost function to be secondary objective. Finally, the comparison among the proposed method and traditional methods are simulated to demonstrate effectiveness and performance.