机器人运动规划方法综述

As application scenarios become more complex, the need for autonomous motion planning techniques which aims at generating collision-free path (trajectory) becomes more urgent. Although a large number of planning algorithms adapted to different scenarios have been proposed already, how to properly classify the existing results and analyze the advantages and disadvantages of different methods is still a problem that needs in-depth consideration. In this paper, the basic connotation of motion planning and the key steps of classical algorithms are explained. Secondly, aiming at the contradiction between real-time performance and the quality of solution path (trajectory), the existing algorithm acceleration strategies are analyzed and summarized hierarchically based on whether differential constraint is considered. Finally, facing the new requirements of planning under uncertainty (i. e., sensor uncertainty, future state uncertainty and environmental uncertainty) and intelligent planning, the latest achievements and development direction in the field of motion planning are reviewed. It is expected that the review can provide ideas for future research.