Article Details

Abstract

To address the issues of insufficient dynamic adaptability of the A* algorithm and the slow convergence and low exploration efficiency of the Deep Q-Network (DQN) algorithm in UAV path planning within complex dynamic scenarios, this study proposes an A*-DQN fusion path planning algorithm. This algorithm utilizes the static global optimal path generated by A* as guidance. It incorporates a hierarchical progressive reward mechanism combined with an adaptive weight adjustment strategy oriented toward dynamic obstacles to achieve a balance between global path optimization and dynamic obstacle avoidance. Experimental results demonstrate that, compared with the traditional DQN algorithm, the A*-DQN algorithm achieves higher convergence speed and test success rates. Furthermore, the efficiency and stability of path planning in dynamic environments are improved. This study provides a feasible technical approach for UAV path planning in dynamic environments and holds practical reference value.

Keywords

A*-DQN; UAV path planning; dynamic obstacle avoidance; reinforcement learning

References

1. [1] Zhao Yinfei, Li Zhongming, Ren Xinhui Overview of UAV Logistics Route Planning Methods in Urban Scenarios [J]. Journal of Nanjing University of Aeronautics and Astronautics (Natural Science Edition), 2025, 57 (6): 1013-1025
2. [2] Huang Xin, Zhang Zhijia, Sun Ping, etc Overview of Path Planning Algorithms Based on Deep Reinforcement Learning [J]. Robotics, 2026, 48 (1): 196-216
3. [3] Tian Xuguang, Zuo Qinwen, Wang Ding, etc Research on Deep Reinforcement Learning Path Planning Algorithm Based on Fusion A * Heuristic Search [J]. Journal of Ordnance Engineering, 2025, 46 (S2): 315-327
4. [4] Li Junlin, Xiong Xingzhong, Yang Kailai, etc Research on Indoor UAV Path Planning Based on ROS and Fusion Algorithm [J]. Foreign Electronic Measurement Technology, 2024, 43 (1): 173-181
5. [5] Wen Haijun, Li Xin Robot Path Planning Based on Improved Fusion DQN Algorithm [J/OL]. Computer Applications and Software, 2026:1-10 https://link.cnki.net/urlid/31.1260.tp.20260108.1646.010.
6. [6] Tang Jianing, Yang Xin, Zhou Sida, etc Research on improving DDQN for unmanned aerial vehicle exploration trajectory planning in unknown environments [J]. Electro Optics and Control, 2023, 30 (4): 23-27, 33
7. [7] Meng Bin, Luna, Fu Shuai, etc Research on Autonomous Navigation and Obstacle Avoidance of Unmanned Aerial Vehicles in Complex Environments Based on Deep Reinforcement Learning [J/OL]. Flight Mechanics (2026-01-26) [2026-02-24] Modified Flight Mechanics, 2026:1-7 https://doi.org/10.13645/j.cnki.f.d.20260123.001.
8. [8] Kou Kai, Yang Gang, Zhang Wenqi, etc Research on Autonomous Navigation Method for Drones Based on SAC [J]. Journal of Northwestern Polytechnical University, 2024, 42 (2): 310-318
9. [9] Lv Maolong, Ding Chenbo, Han Haoran, etc Autonomous perception planning control strategy for unmanned aerial vehicles based on deep reinforcement learning [J]. Journal of Automation, 2025, 51 (6): 1305-1319
10. [10] Liang Xiuman, Liu Ziliang, Liu Zhendong Improved RRT algorithm path planning based on deep reinforcement learning [J]. Journal of System Simulation, 2025, 37 (10): 2578-2593

Funding

Science and Technology Innovation Project Funding for Higher Education Institutions in Shanxi Province (2024L565)

About the authors

Ji Feng (1989-), male, Han ethnicity, from Xinzhou, Shanxi, lecturer, master's student, research direction: machine learning, embedded technology.