A real-time nearly time-optimal point-to-point trajectory planning method using dynamic movement primitives

Abstract

This paper focuses on highly efficient real-time capable time-optimal trajectory planning for point-to-point motions. A method based on the concept of dynamic movement primitives is developed. Thereby an offline generated time-optimal point-to-point trajectory considering nonlinear physical constraints serves as reference for a movement primitive. Through variation of the goal position in a small range around the reference target, new nearly time-optimal point-to-point trajectories are obtained. For the required reconsideration of the physical constraints, a strategy is derived from a common minimum-time optimization problem formulation. Finally a comparison between this and existing realtime capable time-optimal trajectory planning methods is drawn using a six degree-of-freedom serial robot.