复杂环境下基于RRT的智能车辆运动规划算法

RRT-based Motion Planning Algorithm for Intelligent Vehicle in Complex Environments

在存在大量无规则障碍物且障碍物分布不均匀的复杂环境下,现有规划算法不能很好地解决智能车辆的运动规划问题.为此,本文提出了一种简单实用的基于RRT(快速搜索随机树)的运动规划算法——连续曲率RRT算法.该算法在RRT框架中结合了环境约束以及车辆自身的约束.它首先引入了目标偏向采样策略以及合理的度量函数,大大地提高了规划速度和质量;接着提出了一种基于最大曲率约束的后处理方法以生成平滑的且曲率连续的可执行轨迹.通过仿真实验和实车测试,证实了该算法的正确性、有效性和实用性.

The existing planning algorithms can not properly solve the motion planning problem of intelligent vehicle in complex environments with many irregular and random obstacles. To solve the problem, a simple and practical RRT-based algorithm, continuous-curvature RRT algorithm, is proposed. This algorithm combines the environmental constraints and the constraints of intelligent vehicle with RRTs. Firstly, a goal-biased sampling strategy and a reasonable metric function are introduced to greatly increase the planning speed and quality. And then, a post-processing method based on the max- imum curvature constraint is presented to generate a smooth, continuous-curvature and executable trajectory. Simulation experiments and real intelligent vehicle test verify the correctness, validity and practicability of this algorithm.