RRTCAP? - RRT? Controller and Planner - Simultaneous Motion and Planning

摘要

This paper presents a trajectory planning approach, the Rapidly Exploring Random Tree? Controller and Planner (RRTCAP?), combining the planning phase of a RRT?-based algorithm with the execution phase on a real robot. Despite the necessary adaption of RRT? - planner, using Reed-Sheeps curves as local planning method, a trajectory controller is shown which is necessary for compensating deviations introduced by mechanical tolerances and simplified models. The algorithm is implemented on the existing mobile robot Outdoor MERLIN and evaluated in different scenarios using simulations, as well as hardware experiments.