基于DGB_RRT算法的无人机自主航迹规划

An Autonomous Route Planning of UAV Based on DGB_RRT Algorithm

无人机在输电线路走廊空间环境下在线自主进行空间探索,针对传统RRT算法因采样盲目性导致的效率低下的问题,提出了一种基于动态步长-高斯采样-双向快速扩展随机树算法——DGB_RRT算法。首先,采用高斯采样实现启发式采样目标,减少因搜索范围随机性强生成采样点的数量,然后引入贪婪扩展算法和动态矢量步长优化策略确定该采样点的生成方向和增长步长来确定新节点,提高了随机树的扩展速度。其次,使用路径剪枝的策略来剔除冗余采样点,缩短了路径规划的长度,进一步提高了路径搜索的效率,结合Dubins曲线,在满足无人机曲率约束条件下使得航迹平滑处理,实现在障碍区域无人机矫正位姿实现避障飞行。最后,经多次重复对比实验,证实了提出算法不仅规划航迹运行时长缩短了近一半,而且路径长度也缩短了近40%,该方法在速度和效率方面都有明显提升。

UAV(unmanned aerial vehicle)conduct online autonomous spatial exploration in the spatial environment of transmission line corridors.In order to solve the problem of low efficiency of traditional RRT algorithm due to sampling blindness,a DGB_RRT algorithm based on the dynamic step-size-Gaussian sampling——bidirectional fast expanding random tree algorithm is proposed.Firstly,Gaussian sampling is used to achieve heuristic sampling target,which reduces the number of sampling points generated due to the randomness of the search range.Then the greedy expansion algorithm and the dynamic vector step-size optimization strategy are introduced to determine the generation direction and growth step-size of the sampling point to determine the new node,which improves the expansion speed of the random tree.Secondly,a route pruning strategy is used to eliminate redundant sampling points,which shortens the length of route planning and further improves the efficiency of route search.Combined with the Dubins curve,the trajectory is smoothed under the constraint of UAV curvature and the obstacle avoidance flight is achieved by correcting the posture of the UAV in the obstacle area.Finally,through repeated comparative experiments,it is confirmed that the proposed algorithm not only reduces the running time of the planned flight route by nearly 50%,but also reduces the route length by nearly 40%.The proposed algorithm has significant improvement in both speed and efficiency.