Supervisor:
Southwest Ordnance Industry Bureau
Organizer:
Chongqing Ordnance Industry Society
Chongqing University of Technology
Chongqing University of Technology
Adaptive artificial potential track planning method based on reinforcement learning
| DOI: | 10.11809/bqzbgcxb2024.11.004 |
| Keywords: | drone; artificial potential field method; reinforcement learning; path planning; obstacle avoidance; physical constraints |
| Abstract: | In the problem of track planning, the traditional artificial potential field method has difficulty in parameter design and poor scene adaptability, so a track planning method based on reinforcement learning to adjust the artificial potential field parameters is proposed.An artificial potential field deep network model is established, and the network is trained by DDPG reinforcement learning, and the trained deep network adaptive design parameters are used to have better robustness for complex scene adaptability.The simulation results show that the method of using reinforcement learning to design the artificial potential field has a shorter path, better planning effect and stronger scene adaptability, indicating that the method takes advantage of the artificial potential field without search and fast speed.On the other hand, the exploratory characteristics of reinforcement learning are used to greatly improve the efficiency of track planning. |
| Published: | 2024-11-30 |