Intelligent vehicle path planning method based on bidirectional expanded random tree

Abstract

The invention discloses a path planning method for an intelligent vehicle based on a bidirectional expanded random tree, which comprises generating a curve, constructing two expanded trees with opposite starting points when there are obstacles on the curve; generating random points on a map, and searching for the expanded trees The node closest to the random point; determine whether there is an obstacle on the connection between the random point and the node; if so, regenerate the random point; if not, take a new node on the connection between the random point and the node; determine the nearest Whether there is an obstacle on the connection line between the node and the new node, if so, regenerate a random point, otherwise update the node according to the obstacle safety distance constraint; then select the parent node and child node of the node according to the maximum steering constraint of the intelligent vehicle and the node node; judge whether the distance between the two expanded trees is less than the set threshold, if it is greater than, then regenerate a random point, otherwise, according to the two expanded tree nodes, the path from the node to the starting point and the target point is reversely selected to form a smart car planning path .

Description

technical field [0001] The invention relates to the planning of the motion path of an object, in particular to a path planning method for an intelligent vehicle based on a bidirectional expanded random tree. Background technique [0002] In 2001, Kuffner and LaValle proposed a method of generating random trees in parallel (RRT-Connect) to improve the convergence speed of the algorithm. In 2010, Karaman and Frazzoli proposed the RRT* algorithm to solve the problem that the path generated by the RRT algorithm is not a probabilistic optimal solution, but its operation efficiency is low. In 2013, M.Jordan and A.Perez proposed B-RRT*, using the RRT* algorithm to expand the starting point and the target point in both directions to reduce the execution time of the RRT* algorithm, which is applied to mobile robots, B- Although RRT* can shorten part of the execution time, it still has the following disadvantages: [0003] (1) During the running process, global uniform random sampli...

AI Technical Summary

Problems solved by technology

[0003] (1) During the running process, global uniform random sampling is used, which increases the running time of the algorithm and slows down the convergence speed;

[0004] (2) Using the nearest node selection algorithm for node selection can easily lead to the failure of the planned path in complex scenarios;

[0005] (3) The planned path does not consider the vehicle kinematics constraints, so that the planned path cannot be used in the path planning of intelligent vehicles

Images