TY - GEN
T1 - Path Planning Algorithm for Multiple UAVs Based on Artificial Potential Field
AU - Ren, Chongde
AU - Chen, Jinchao
AU - Du, Chenglie
AU - Yu, Wenquan
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Due to the lower cost and higher maneuverability, unmanned aerial vehicles (UAVs) have found extensive use in both the civilian and military worlds. Path planning, as a crucial problem in the process of UAVs flight, aims to determine the optimal routes for multiple UAVs from various starting points to a single destination. However, because of the involvement of complex conditional constraints, path planning becomes a highly challenging problem. The path planning problem involving numerous UAVs is examined in this research, and a SAAPF-MADDPG algorithm based on Artificial Potential Field (APF) is suggested as a solution. First, a SA-greedy algorithm that can change the probability of random exploration by agents based on the number of steps and successful rounds to prevent UAVs from getting trapped in a local optimum. Then, we design complex reward functions based on APF to guide UAVs to destination faster. Finally, SAAPF-MADDPG is evaluated against the MADDPG, DDPG, and MATD3 methods in simulation scenarios to confirm its efficacy.
AB - Due to the lower cost and higher maneuverability, unmanned aerial vehicles (UAVs) have found extensive use in both the civilian and military worlds. Path planning, as a crucial problem in the process of UAVs flight, aims to determine the optimal routes for multiple UAVs from various starting points to a single destination. However, because of the involvement of complex conditional constraints, path planning becomes a highly challenging problem. The path planning problem involving numerous UAVs is examined in this research, and a SAAPF-MADDPG algorithm based on Artificial Potential Field (APF) is suggested as a solution. First, a SA-greedy algorithm that can change the probability of random exploration by agents based on the number of steps and successful rounds to prevent UAVs from getting trapped in a local optimum. Then, we design complex reward functions based on APF to guide UAVs to destination faster. Finally, SAAPF-MADDPG is evaluated against the MADDPG, DDPG, and MATD3 methods in simulation scenarios to confirm its efficacy.
KW - autonomous navigation
KW - deep reinforcement learning
KW - multi-agent systems
KW - path planning
UR - http://www.scopus.com/inward/record.url?scp=85186097142&partnerID=8YFLogxK
U2 - 10.1109/ITAIC58329.2023.10408816
DO - 10.1109/ITAIC58329.2023.10408816
M3 - 会议稿件
AN - SCOPUS:85186097142
T3 - IEEE Joint International Information Technology and Artificial Intelligence Conference (ITAIC)
SP - 970
EP - 974
BT - IEEE ITAIC 2023 - IEEE 11th Joint International Information Technology and Artificial Intelligence Conference
A2 - Xu, Bing
A2 - Mou, Kefen
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 11th Joint International Information Technology and Artificial Intelligence Conference, ITAIC 2023
Y2 - 8 December 2023 through 10 December 2023
ER -