Abstract
Understanding the adaptive exit choice behaviours during pedestrian evacuation, along with the associated complex phenomena, is essential for developing effective evacuation models and ensuring pedestrian safety. In this study, we conducted a series of virtual evacuation experiments involving multiple simultaneous participants under normal and fire conditions. The first-person perspective videos, trajectories, and rotation data of visual centre points were collected to analyse pedestrians’ microscopic behaviours and exit choices over time. The results showed that pedestrians exhibited patterns of multi-attribute conjoint decision-making, with obstacles and fires influencing movement and exit choices. Additionally, initial decision points of pedestrians were mainly located near the entrance, and decision changes were observed. However, such changes were infrequent. The more hazardous the condition, the less likely pedestrians were to alter their initial exit choices. A post-experiment survey was designed to assess participants’ perceptions and exit choice strategies during evacuation. The results indicated that the virtual environment closely resembled reality and fire emergencies significantly impacted exit choice strategies. Pedestrians primarily considered factors including distance to exits, crowd density near exits, and risks of exits in their decision-making process, and the distance was the most influential factor in all scenarios. Finally, a multinomial logit model was developed and calibrated for exit choice prediction using these three factors. This model was integrated into an extended multi-grid cellular automata model for evacuation simulation, effectively replicating pedestrian movement and adaptive exit choice behaviours with minor differences. These findings provide valuable theoretical insights and simulation support for future building safety assessments. © 2025 Elsevier Ltd.
Original language | English |
---|---|
Article number | 103302 |
Journal | Advanced Engineering Informatics |
Volume | 65 |
Issue number | Part C |
Online published | 4 Apr 2025 |
DOIs | |
Publication status | Published - May 2025 |
Funding
This study was supported by the National Natural Science Foundation of China (No. 72104254), the Hubei Provincial Natural Science Foundation of China (No. 2022CFB469), and the Research Start-up Foundation of South-Central Minzu University (YZZ21002). We are grateful to Key Laboratory of Cyber-Physical Fusion Intelligent Computing (South-Central Minzu University) and Hubei Provincial Engineering Research Center for Intelligent Management of Manufacturing Enterprises for the laboratory aspects of this study.
Research Keywords
- Cellular automata
- Decision adaptation
- Evacuation dynamics
- Exit choice model
- Survey
- Virtual experiment