Scholarlink

<p>In robotics, reinforcement learning (RL) has advanced autonomous navigation significantly. However, protecting privacy and security in cloud-based autonomous systems remains a major challenge. In this paper, we introduce a cloud-based framework for secure and privacy-preserving robotic path planning by leveraging the power of deep reinforcement learning. Our method combines DQN and PPO with selective homomorphic encryption and differential privacy to secure sensitive data while at rest, in transit, and during computation. We empirically show that our method achieves better path efficiency, computational expense, and security properties than relevant baselines. Our approach leverages deep reinforcement learning (DRL) algorithms for efficient decision-making with privacy-preserving methods such as differential privacy and homomorphic encryption. Tests demonstrate the effectiveness of our approach in achieving high-performance robotic route planning while lowering privacy and security risks. The findings indicate a 27% gain in security robustness, a 23% increase in route efficiency, and an 18% reduction in computation cost when compared to traditional methods.</p>

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

Privacy preservation and security for the execution of cloud-based optimized robot path using reinforcement learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月24日

查看全文
  • Revati Raman Dewangan,
  • Sunita Soni,
  • Deepali Thombre,
  • Monika Arya,
  • Bhupesh Kumar Dewangan

摘要

In robotics, reinforcement learning (RL) has advanced autonomous navigation significantly. However, protecting privacy and security in cloud-based autonomous systems remains a major challenge. In this paper, we introduce a cloud-based framework for secure and privacy-preserving robotic path planning by leveraging the power of deep reinforcement learning. Our method combines DQN and PPO with selective homomorphic encryption and differential privacy to secure sensitive data while at rest, in transit, and during computation. We empirically show that our method achieves better path efficiency, computational expense, and security properties than relevant baselines. Our approach leverages deep reinforcement learning (DRL) algorithms for efficient decision-making with privacy-preserving methods such as differential privacy and homomorphic encryption. Tests demonstrate the effectiveness of our approach in achieving high-performance robotic route planning while lowering privacy and security risks. The findings indicate a 27% gain in security robustness, a 23% increase in route efficiency, and an 18% reduction in computation cost when compared to traditional methods.