Scholarlink

The coverage problem is to visit as many possible points with less overlap in an environment that may be unknown. This problem is known to be NP-hard. We propose a multi-agent deep reinforcement learning approach for the coverage problem, ensuring reduced overlap. We introduce a novel time variant reward function that encourages agents to cover the environment more efficiently with less overlap. In our approach, a coverage improvement of 43.94% to 88.96% is obtained compared to existing methods. There is a maximum overlap reduction of 51.80%–67.5% and an average overlap reduction of 64.40%–78.3% compared to some existing approaches.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-agent Deep Reinforcement Learning for Coverage in Unknown Environments

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nirali Sanghvi,
  • Rajdeep Niyogi,
  • Onika Yadav

摘要

The coverage problem is to visit as many possible points with less overlap in an environment that may be unknown. This problem is known to be NP-hard. We propose a multi-agent deep reinforcement learning approach for the coverage problem, ensuring reduced overlap. We introduce a novel time variant reward function that encourages agents to cover the environment more efficiently with less overlap. In our approach, a coverage improvement of 43.94% to 88.96% is obtained compared to existing methods. There is a maximum overlap reduction of 51.80%–67.5% and an average overlap reduction of 64.40%–78.3% compared to some existing approaches.