High Precision Trajectory Learning Method Based Improved Dynamic Movement Primitives for Robot Skill Learning

Abstract

Trajectory learning is an important part of robot skill learning, and a trajectory learning method based on improved Dynamic Movement Primitives (DMPs) is proposed to improve trajectory reproduction accuracy. In this method, the truncation processing is used to improve the Gaussian kernel function of DMPs to eliminate the impact of tail exponential decay on fitted target forcing term, and the optimization on the number of shape parameters is used to make the model better approximate the local gradient of the target forcing term. The principle of trajectory accuracy improvement is described in detail. The trajectory reproduction simulation is performed, which verifies the feasibility of the proposed method. An experimental setup for robot skill trajectory learning is constructed and the relevant comparison experiments are performed, which verifies the effectiveness of the proposed method in improving trajectory learning accuracy.