Scholarlink

Achieving realistic crowd simulation in Virtual Reality (VR) is crucial for applications such as training, emergency preparedness, and AI - driven crowd modeling. A key aspect of these simulations is understanding how individuals make decisions under various conditions. This work focuses on emergency scenarios. Specifically, this study investigates the influence of virtual guides (artificial agents), environmental familiarity, and crowd patterns on evacuation decision-making. Findings reveal that evacuees do not blindly follow artificial agents but integrate spatial and social cues when selecting routes. Compliance is influenced by the alignment between guide instructions, crowd movement, and prior spatial knowledge, with uneven crowd distributions exerting a stronger effect than familiarity. These insights contribute to the development of adaptive virtual agents, AI-enhanced crowd simulations, and VR-based evacuation training, supporting data-driven emergency response systems.

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

Virtual Guides and Crowd Behaviors: Understanding Evacuation Decision-Making in Virtual Reality

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ruochen Cao,
  • Ziyuan Feng,
  • Changyue Ma,
  • Xin Wen,
  • Yanrong Hao,
  • Chenchen Zhang,
  • Zequn Liang,
  • Ziarmal Hussain,
  • Rui Cao

摘要

Achieving realistic crowd simulation in Virtual Reality (VR) is crucial for applications such as training, emergency preparedness, and AI - driven crowd modeling. A key aspect of these simulations is understanding how individuals make decisions under various conditions. This work focuses on emergency scenarios. Specifically, this study investigates the influence of virtual guides (artificial agents), environmental familiarity, and crowd patterns on evacuation decision-making. Findings reveal that evacuees do not blindly follow artificial agents but integrate spatial and social cues when selecting routes. Compliance is influenced by the alignment between guide instructions, crowd movement, and prior spatial knowledge, with uneven crowd distributions exerting a stronger effect than familiarity. These insights contribute to the development of adaptive virtual agents, AI-enhanced crowd simulations, and VR-based evacuation training, supporting data-driven emergency response systems.