基于启发式Q学习的输电线路无人机避障导航巡检

The obstacle avoidance navigation inspection of unmanned aerial vehicles based on heuristic Q-learning

为提高无人机在输电线路巡检中的安全性,设计一种基于启发式Q学习算法的输电线路无人机自动避障巡检方法。运用多传感器采集外界环境变化信息,经信息融合获得有价值的路径感知信息,得到障碍物位置和可通行路径;采用Q学习算法与启发式函数计算信息强度,提前预测分析飞行动作的重要程度,得出最佳飞行动作和位置,完成巡检中的无人机自动避障导航。实验结果表明:在静态环境中,所提方法在训练次数较少的情况下达到收敛,不仅能成功绕过静态与动态障碍物,且飞行避障导航路径短。

In order to improve the security of UAV in transmission line inspection,an automatic obstacle avoidance inspection method is designed based on heuristic Q learning algorithm.Multi sensors are used to collect the change information of the external environment,and valuable path perception information is obtained through information fusion to obtain the location of obstacles and accessible paths;Q learning algorithm and heuristic function are used to calculate information intensity,predict and analyze the importance of flight action in advance,give the best flight action and position,and complete the automatic obstacle avoidance navigation of UAV in patrol inspection.The experimental results show that the proposed method converges in a static environment with less training times.It can not only successfully bypass static and dynamic obstacles,but also have a short flight obstacle avoidance navigation path.