Scholarlink

In this paper, we introduce an efficient group-aware social robot navigation model, GNav. The GNav model is specifically designed to explicitly model group interactions utilizing a spatial-temporal interaction graph. This design choice significantly enhances the model’s ability to navigate environments that contain a mixture of groups and individual pedestrians. Evaluations conducted in simulated settings indicate that our model outperforms the performance of current state-of-the-art approaches in terms of the success rate and the group intersection rate, highlighting its effectiveness and adaptability in diverse social navigation contexts.

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

Socially Aware Mobile Robot Navigation: A Deep Reinforcement Learning-Based Group-Aware Model

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Minh Hoang Dang,
  • Viet-Binh Do,
  • Doan Manh Dung,
  • Nghiem Hoang Nam,
  • Xuan-Tung Truong

摘要

In this paper, we introduce an efficient group-aware social robot navigation model, GNav. The GNav model is specifically designed to explicitly model group interactions utilizing a spatial-temporal interaction graph. This design choice significantly enhances the model’s ability to navigate environments that contain a mixture of groups and individual pedestrians. Evaluations conducted in simulated settings indicate that our model outperforms the performance of current state-of-the-art approaches in terms of the success rate and the group intersection rate, highlighting its effectiveness and adaptability in diverse social navigation contexts.