Work

Multi-Agent Target Pursuit Using Perception Uncertainty-Aware Reinforcement Learning

Year: 2024

Type: conference-paper

Abstract: Existing target pursuit systems are able to coordinate a team of mobile agents to capture or intercept unauthorized targets. Multi-agent reinforcement learning (MARL) further empowers pursuit strategies with the potential to emerge complex behaviors. However, existing solutions lack the ability to handle the perception uncertainty caused by relative position measurement noises, which blurs the understanding of the target's state and complicates the pursuit strategy learning process. This study proposes PUARL, which enhances the learning under the perception uncertainty process by guiding exploration with probabilistic estimation and adapting the policy based on awareness of perception uncertainty. We validate its performance in terms of both accuracy and efficiency. PUARL achieves a success rate increase of 12.3%+ and a reduction in total steps by 58.3%+, outperforming both state-of-the-art heuristic and learning-based solutions.

Source: Proceedings of the 28th Annual International Conference on Mobile Computing And Networking

DOI: 10.1145/3636534.3694724

Authors: Yuhan Cheng, Jirong Zha, Renjue Yang, Zhi Sun, Susu Xu, Xinlei Chen

Institutions: Tsinghua Shenzhen International Graduate School, Tsinghua University, Johns Hopkins University, Peng Cheng Laboratory

Language: English

Topic: Reinforcement Learning in Robotics

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

OpenAlex ID: W4405014268

FWCI: 2.647

Cites: 33

Cited by: 9

Related to: 10