Research On Path Planning Of Unmanned Vehicles Based On Rapidly-exploring Random Tree

The path planning of unmanned vehicle refers to the planning of a path that does not collide with the obstacles given the starting position,ending position of the vehicle and the distribution of obstacles in the environment.In recent years,new path planning algorithms have continuously appeared and developed.The most representative and common path planning algorithms in the field of path planning are mainly divided into map-based path planning algorithms,bionics-based path planning algorithms,and sampling-based path planning algorithm.The sampling-based search algorithms include the probability graph algorithm and Rapidly-exploring Random tree algorithm.The advantage of the Rapidly-exploring Random tree algorithm is that it does not need to model the planned space.It is a random sampling algorithm and considers the constraints of the objective existence of unmanned vehicles,so it is widely used.The basic Rapidly-exploring Random tree algorithm also has the following shortcomings in path planning:(1)the path is randomly generated,the path is biased;(2)the random tree is no oriented in the search process;(3)the convergence speed is slow,and the search efficiency is low.In this paper,the Rapidly-exploring Random tree algorithm is taken as the research object.Based on the analysis of the Rapidly-exploring Random tree algorithm,the basic Rapidly-exploring Random tree algorithm is improved by directivity and heuristic function to realize the vehicle path optimization.Based on the kinematic analysis of the vehicle,the vehicle coordinate system and the vehicle moving steering model are established.Through the steering constraints,the planning path can meet the driving requirements of the vehicle,which lays the foundation for the following path planning experimental research.In view of the shortcomings of the Rapidly-exploring Random tree algorithm,bidirectional random trees and multiple local random trees are used to explore and merge.By increasing the gravitational component,the bidirectional random trees grow towards their respective targets,which reduces the randomness of the algorithm;based on the root node generated around the obstacle,the disturbance of repulsion force is added to the root node,and multiple local random trees are generated to quickly find the path that can pass,reduce the detection time of the obstacle in the expansion process,accelerate the convergence speed of the algorithm,and improve the deviation of the algorithm.This paper studies and analyzes the generation method of random points in the Rapidly-exploring Random tree algorithm,and uses the search method of cyclic alternating iteration and bidirectional random tree cooperation to generate new nodes,further optimizes the basic Rapidly-exploring Random tree algorithm,and reduces the number of invalid sampling points.In order to improve the quality of path planning,the generated path is optimized to remove redundant nodes,reducing the number of nodes,shortening the length of the path,and improving the effectiveness of the path.The B-spline curve is used to insert local endpoints to smooth the path,so that the generated path is more in line with the driving conditions of the vehicle,and the quality of the path is improved.Through the CarSim / Matlab joint simulation test,the generated path is tracked,and the effectiveness of the improved algorithm is verified.The generated path meets the driving requirements of the vehicle.Build the ROS intelligent vehicle test platform,transplant the improved algorithm to the intelligent vehicle,and verify the feasibility of the algorithm through trajectory planning.