Research on Intelligent Vehicle Path Planning Method Based on Improved RRT Algorithm

Abstract

This paper studies the problem of intelligent vehicle path planning based on improved RRT algorithm. Firstly, the extended target bias strategy is used to make the random point sampling of RRT algorithm biased to the target point. Secondly, based on the vehicle kinematics model, the random point sampling area is limited to the feasible fan-shaped range in front of the vehicle that is greater than the step size d. Thirdly, in the process of random point expansion, considering the actual size of vehicles and obstacles, the separation axis law is used to detect the collision risk of vehicles and surrounding obstacles in real time. Finally, based on the Matlab simulation platform, different RRT algorithms are used for simulation and analysis in different environments, and the improved RRT path planning algorithm is verified on the ROS intelligent vehicle platform. The experimental results show that compared with the basic RRT algorithm, the improved RRT algorithm reduces the average number of sampling points by 51. 8 percent, and the search time is reduced by 41. 7 percent in the right-angle turn road environment. In the s-shaped road environment, the average number of tree nodes is reduced by 54.3 percent, and the search time is reduced by 42 percent. In the maze-type road environment, the average number of tree nodes is reduced by 70 percent, and the search time is reduced by 52.2 percent. In the dynamic obstacle environment, the real-time performance of dynamic obstacle avoidance is good and the path is smooth. This paper considers the improved RRT algorithm of vehicle kinematics and geometric size constraints. The search efficiency and smoothness are improved, and the generated path meets the requirements of intelligent vehicle path tracking.