A velocity-domain MAPPO approach for perimeter defensive confrontation by USV groups

Open Webpage

Huayan Pu, Jinduo Wang, Senhui Gao, Zhaoxiang Shi, Qun Deng, Yangmin Xie

Published: 01 Jan 2025, Last Modified: 12 Apr 2025Expert Syst. Appl. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Proposal of RAV for robust action reward evaluation in RL frameworks.•Addressing sparse rewards: Velocity-domain advantage estimation.•Outperforms distance-based RL in diverse confrontational PDC scenarios.