Merging Graph Theory and VR: Human-Centred Assessment of Fire Evacuation Cognition

Conventional fire safety analysis focuses on distances and capacities but offers little insight into how occupants perceive and understand complex layouts under stress. This paper proposes a human-centred computational framework that links building spatial configuration with evacuation cognition. We introduce the Evacuation Cognitive Index (ECI), a composite metric that integrates Spatial Complexity (SC) from Graph Theory with Visual Load (VL) captured through eye tracking in immersive Virtual Reality (VR) evacuation scenarios. A BIM-based model provides the geometric and topological basis for GT analysis, while VR simulations record participants’ gaze behaviour, route choice, and evacuation time in a school building under simulated fire conditions. Higher ECI values are associated with denser fixation patterns, increased decision uncertainty, and longer evacuation times, whereas low ECI routes support clearer spatial legibility. ECI thus functions as a cognition-informed design indicator that advances evacuation planning toward humanistic, perception-aware and performance-based building safety.