Path planning algorithm for autonomous driving based on improved RRT*
Wenhao Wu, Yuhai Gu, Yifan Zheng · 2025
Path planning is an important link in the research of autonomous vehicles. Based on a given environment model and under certain constraints, path planning plans a collision-free path connecting the current location of the vehicle and the target location, and selects according to different requirements to obtain optimal path. With the continuous development of various fields, the traditional path planning algorithm has problems such as slow planning speed and large amount of data calculation under the complex environment model, and it is difficult to meet the requirements of real-time and accuracy of path planning. Therefore, this paper builds an intelligent driving platform based on Autoware.Universe, which can realize real-time positioning, path planning, autonomous obstacle avoidance and other functions. The article adopts the improved RRT* path planning algorithm, introduces self-heuristic thinking to constrain the generation of sampling points, prunes redundant nodes in the path, and then uses the Elastic band method to smooth the path. The path planning efficiency is increased by 45.7%. Finally, import the Autoware planning module, and use the ROS2 communication mechanism to send CAN messages to the chassis VCU to realize the global path planning of the vehicle. The test results show that the improved algorithm can effectively realize autonomous obstacle avoidance and planning functions, and the planning effect achieved is more effective.