智能车辆避障路径规划方法研究

Obstacle avoidance path planning approaches of autonomous vehicle

为解决车辆在避障时由于路径曲率不连续易发生原地转向的问题,该文对比探究智能车辆避障路径规划方法,开展了五次贝塞尔曲线、三次贝塞尔曲线、五次样条曲线和三次样条曲线避障路径研究。首先,确定控制点分别拟合出五次贝塞尔曲线、三次贝塞尔曲线、五次样条曲线和三次样条曲线的避障路径;其次,采用Matlab和Carsim联合仿真分析的方法对该四种避障路径的曲率进行对比分析;最后,基于无人驾驶平台,采用纯跟踪预瞄模型开展实车试验,以此探究四种避障路径的工程实用性。结果表明:采用样条曲线和贝塞尔曲线进行避障规划,车辆的横向超调量为0;相比于三次曲线避障方法,五次曲线避障方法所规划的路径曲率更连续;相比于样条曲线避障方法,贝塞尔曲线避障方法车辆的转向更稳定。

To resolve the issue that vehicles tend to turn in place due to discontinuous path curvature under the condition of avoiding obstacles,the obstacle avoidance path planning methods of intelligent vehicles are discussed and quintic Bezier curve,cubic Bezier curve,quintic spline curve and cubic spline curve are adopted to conduct local path planning.Firstly,obstacle avoidance of the four curves are fit with the adoption of control points.Secondly,curvatures of the four obstacle avoidance paths are analyzed by employing Matlab and Carsim.Eventually,real vehicle test based on tracked platform is carried out to explore engineering practicability of the four obstacle avoidance paths by utilizing Pure Pursuit model.The results indicate that vehicle lateral overshoot is 0 when adopting spline curve and Bezier curve for obstacle avoidance planning.Moreover,the path curvature of quintic curve obstacle avoidance method is more continuous than that of cubic curve obstacle avoidance method,and vehicle is possessed with more stable steering under Bezier curve obstacle avoidance method than spline curve obstacle avoidance method.