Scholarlink

This paper introduces Multi-Agent Reinforcement Learning with Proximal Policy Optimization (MARLPPO), an innovative framework for multi-agent drone surveillance using a 3D Unity simulation and Deep Reinforcement Learning (DRL). This approach utilizes dynamic policy adjustments and agent collaboration to minimize redundancy and enhance monitoring. MARLPPO offers a scalable and efficient solution for drone-based real-world surveillance applications, demonstrating superior performance metrics in experimental results. By integrating Proximal Policy Optimization (PPO) with Multi-Agent Reinforcement Learning (MARL), MARLPPO reduces surveillance time by up to 40% compared to traditional models and increases area coverage by 25%.

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

MARLPPO: Multi-agent Drone Surveillance Using Deep Reinforcement Learning in Urban Aerial Mobility

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Gopi Banavathu,
  • Viswesh Nanapu,
  • K. E. Srinivasa Desikan

摘要

This paper introduces Multi-Agent Reinforcement Learning with Proximal Policy Optimization (MARLPPO), an innovative framework for multi-agent drone surveillance using a 3D Unity simulation and Deep Reinforcement Learning (DRL). This approach utilizes dynamic policy adjustments and agent collaboration to minimize redundancy and enhance monitoring. MARLPPO offers a scalable and efficient solution for drone-based real-world surveillance applications, demonstrating superior performance metrics in experimental results. By integrating Proximal Policy Optimization (PPO) with Multi-Agent Reinforcement Learning (MARL), MARLPPO reduces surveillance time by up to 40% compared to traditional models and increases area coverage by 25%.