Research for UAV Path Planning Method Based on Guided Sarsa Algorithm
He Boming, Wei Lin, Mei Fuzeng, Huahao Fan · 2022
As unmanned aerial vehicles (UA V) are widely used in unknown and complex environments, their path planning capabilities face higher requirements. In many cases, UA V cannot obtain the environmental information of the target area in advance, and the reinforcement learning (RL) algorithm to solve this problem faces the problem of slow convergence speed. To solve this problem, a UA V path planning method based on guided Sarsa algorithm is proposed, which defines the return function based on position information and improves the status update strategy. Simulation results show that the proposed method can realize fast path planning of UA V in static environment. Compared with Q learning and unimproved Sarsa algorithm, the obtained path length is shortened by 18 steps and 2 steps. At the same time, after 2000 iterations, only the results obtained by this method have obvious convergence. The convergence speed of guided Sarsa algorithm is relatively accelerated. At the same time, the trajectory tracking process of UA V is realized by MAT LAB platform, which proves the practicability and feasibility of the algorithm.