For the issue of traditional unmanned aerial vehicle (UAV) local path planning algorithms easily falling into local optima, an algorithm for local path planning that combines deep reinforcement learning and memory-guided strategy is proposed. The algorithm takes the UAV heading angle as the learning action, and detection information as the state. It integrates the Deep Q-network (DQN) algorithm to learn path planning strategies. When the UAV gets trapped in a local optimum, the memory-guided strategy utilizes historical detection information to guide the UAV out of the local optimum, completing local path planning for UAVs in unknown environments. Simulation experiments verify that the proposed algorithm achieves higher success rates and shorter average flight distances.
Deep Reinforcement Learning-Based Local Path Planning with Memory-Guided
Lect. Notes Electrical Eng.
International Conference on Guidance, Navigation and Control ; 2024 ; Changsha, China August 09, 2024 - August 11, 2024
2025-03-11
10 pages
Article/Chapter (Book)
Electronic Resource
English
Unmanned Aerial Vehicles Path Planning Based on Deep Reinforcement Learning
Springer Verlag | 2019
|UAV path planning with kinematic constraints based on deep reinforcement learning
British Library Conference Proceedings | 2022
|