基于多影响因素RDMA*算法的无人驾驶动态路径规划

Dynamic Path Planning for Unmanned Driving Based on Multi-influencing Factors RDMA* Algorithm

无人驾驶汽车的路径规划面临着复杂多变的交通环境,为了更全面的评价路径选择指标以规划更合理的路径,以及更好的解决路段环境动态变化对规划结果造成的影响,研究了一种考虑多影响因素的动态路径规划算法——RDMA*(Real-time Dynamics of Multiple influencing factors AStar)算法。以A*(AStar)算法为核心,通过加入多影响因素的交通评价因子对其代价函数进行改进,综合考虑距离,交通拥堵程度,道路平整度和其他影响因素,应用层次分析法确定各影响因素的相对权重,以综合代价值为评价指标进行路径规划。通过GPS,雷达和摄像头等设备,利用融合感知技术获取相关道路环境信息,根据获取的全局和局部交通环境数据信息,利用实时动态更新策略解决动态环境下的路径规划问题,实时规划最优路径。通过对实际案例进行模拟,结果表明,应用RDMA*方法规划的路径相比基础A*方法规划的路径出行总体耗时减少了15.75%。并且在遇到特殊事件的状况下,通过RDMA*动态规划可为无人驾驶车辆即时提供一条综合代价值最小,耗时最少的可行路径,与改进的A*动态路径规划方法相比减少了10.63%的二次规划综合代价值的损耗,提高了7.83%的时间效率。该方法能更好的适应复杂的道路和交通系统,即时应对动态变化的交通状况,具备更强的实用性。

Path planning of unmanned driving cars faces a complex and changeable traffic environment. In order to more comprehensively evaluate indicators of path selection to plan a more reasonable path and reduce impacts of changes in the road environment on the planning results,a dynamic path planning algorithm of RDMA*(Real-time Dynamics of Multiple influencing factors AStar)is studied. Based on the A*(Astar)algorithm,the cost function is improved by adding traffic evaluation factors with multiple influencing factors,considering the distance,traffic congestion, road smoothness,and other influencing factors,applying AHP to determine the relative weight of each influencing factor,and using the comprehensive cost value as the evaluation index for path planning. GPS, radar and cameras,and other devices are used,and fusion sensing technology is applied to obtain relevant road environment information. According to the obtained global and local traffic environment information,real-time dynamic update strategy is used to solve the path planning problem in dynamic environment and plan the optimal path in real time. By simulating actual traffic cases,the results show that compared to the route planned by the basic A* method, the overall travel time is reduced by15.75%for the path planned by the RDMA* method. And in the case of special events,RDMA*dynamic planning can instantly provide a feasible path with the lowest comprehensive cost value and the least amount of time spent for unmanned driving vehicle. Compared to the improved A* dynamic path planning method,the loss of comprehensive cost value in the secondary planning is reduced by10.63%,and the time efficiency is improved by7.83%. The method can better adapt to complex road and traffic systems, and respond to dynamic changing traffic conditions immediately,has better practicability.