一种无人机路径规划强化学习算法

A reinforcement learning algorithm for combat UAV path planning

为了让无人机不易遭到地面埋伏的单兵防空武器打击,提出了一种新的强化学习算法,用于无人机(UAV)执行规避导弹、最短路径飞行和编队飞行任务.该算法结合自我模仿学习和随机网络提炼算法,以放大探索的模仿效应(AIE).实验结果表明,所提出的算法在寻找UAV最短飞行路径的同时避开敌方导弹方面非常有效;在收敛速度和学习稳定性方面都优于现有算法.这为UAV躲避导弹被击中的事件提供了一定的参考.

In order to make it uneasy for unmanned aerial vehicles(UAVs)to be attacked by the ground ambush of individual anti-aircraft weapons,this paper proposes a new reinforcement learning algorithm used for combat UAVs to perform the mission of missile avoidance,shortest path flight and formation flight.The algorithm combines self-imitation learning and stochastic network refining algorithm to enhance exploration through amplification of imitation effect(AIE).Experimental results show that the proposed algorithm is very effective in finding the shortest flight path for the combat UAV while avoiding enemy missiles,and is also superior to the existing algorithm in terms of convergence speed and learning stability.This provides a certain reference for the UAVs to avoid being hit by missiles.