An Improved RRT* Path Planning Algorithm in Dynamic Environment

Jianyu Li, Kezhi Wang, Zonghai Chen, Jikai Wang

Abstract: With the increasingly diverse usage scenarios of mobile robots, the path planning of mobile robots in dynamic environments has become a hot issue. Aiming at the slow planning speed of RRT* algorithm in dynamic environment, this paper proposes an improved RRT* algorithm: efficient dynamic rapidly-exploring random tree star (ED-RRT*). In a static environment, the algorithm uses goal-biased sampling to reduce the randomness of the RRT* sampling method, and sets the sampling step size to be adaptive to improve the planning speed of the initial path. When the dynamic environment encounters obstacles and needs to be re-planned, the local target points are given based on the initial path and environmental information to quickly complete the local re-planning. The path information planned in the static environment is used as the prior information to improve the efficiency of the replanning algorithm. Finally, in the simulation environment based on OpenCV2, the ED-RRT* algorithm is compared with the improved RRT* algorithm. The experimental results show that ED-RRT* has faster dynamic performance and fewer nodes.