behavioral evacuation modeling

This study addresses the limitations of existing evacuation models, which often assume rational and homogeneous human behavior, thereby failing to capture the complex cognitive, emotional, and social dynamics observed during disasters, compounded by a scarcity of real-world data. To overcome these challenges, this work proposes a novel simulation framework that integrates personality-driven large language models with a three-layer cognitive architecture—comprising goal setting, path reasoning, and low-level navigation—to model heterogeneous and irrational sequential decision-making in dynamic, grid-based disaster environments. Calibrated against empirical data, the proposed approach significantly enhances the realism and predictive accuracy of evacuation simulations, mitigating the overly optimistic biases inherent in conventional models and offering more reliable decision support for emergency planning.

This study addresses the limitation of traditional fixed-rule models in capturing internal human decision-making processes, such as perception and memory, under mobile threats. To this end, it proposes a large language model (LLM)-driven agent evacuation framework. The approach integrates private symbolic views, memory graphs, and persona-based prompting to simulate individual cognition. Furthermore, it introduces a novel paradigm comprising state-compressed context and a validation engine that decouples behavioral selection from physical feasibility, enabling auditable, end-to-end heterogeneous behavior generation. Experimental results demonstrate a strong correlation between knowledge of available exits and evacuation success rates (89.5% vs. 1.05%), revealing the critical influence of personality traits and information acquisition on hazard assessment and evacuation outcomes.

This study investigates the life-cycle evolution of pedestrian groups as emergent intelligent agents and its influence on walking behavior, specifically addressing the relationship between dynamic group cohesion and collective intention formation. Method: We propose a state-transition-based group life-cycle model and develop a data-driven group morphology–behavior coupling model by integrating trajectory extraction from surveillance videos, expert-annotated group relationships and events, and quantitative clustering analysis. Contribution/Results: We systematically identify, for the first time, empirical correlations among cohesion dynamics, enhanced agentivity, and morphological state transitions; further, we derive generalizable abstract walking pattern sequences characterizing group locomotion. These findings provide computationally tractable, interpretable, and structurally grounded behavioral primitives—enabling high-fidelity group simulation modeling and principled design of human–machine collaborative interaction systems.

Existing evacuation models commonly neglect pedestrian collisions, social interactions, and individual heterogeneity—including terrain variations and anthropometric differences—leading to unrealistic simulations. To address these limitations, this paper proposes a parallel multi-agent 3D evacuation simulation framework grounded in a Sense–Decide–Move (SDM) paradigm. The method integrates a 3D adaptive social force model with personalized gait control, enabling accurate modeling of non-planar terrain and dynamic group-level perception. A novel contribution is the introduction of part-level force field visualization, facilitating fine-grained behavioral attribution analysis. Experimental results demonstrate that the framework significantly improves trajectory plausibility and individual behavioral fidelity, while supporting real-time, interpretable dynamic path planning. Compared to conventional models, it achieves superior realism and practical applicability in complex evacuation scenarios.

This study addresses the challenge of modeling human responses to stressors during emergency evacuations in multi-compartment buildings. Methodologically, it introduces a novel social-force model framework integrating psychological stress theory—specifically embedding stress-response mechanisms systematically into macroscopic pedestrian dynamics for the first time—coupled with opinion dynamics to capture collective decision-making during pre-movement phases, and synergistically combining FDS+EVAC with the custom crowdEgress platform for multi-scale co-simulation. Its key contribution lies in establishing an interpretable, causal chain from environmental stress → individual stress response → emergent collective behavior, unifying bottleneck passage, herding decisions, and dynamic pathfinding within a single formalism. Experimental validation demonstrates high-fidelity reproduction and prediction of stress-induced phenomena—including evacuation delays, localized congestion amplification, and irrational clustering—as well as accurate reconstruction of real-world flow-rate distributions and route-choice preferences across benchmark scenarios.

This study addresses pedestrian-friendly infrastructure design by developing a microscopic pedestrian flow dynamics model grounded in social force theory, which captures self-organized behaviors emerging from individuals’ goal-directed motion and avoidance of obstacles and other pedestrians. The work innovatively integrates evolutionary algorithms to optimize building layouts and introduces mechanisms for dynamic target selection, experience-based learning, and a path load distribution method that accounts for subjective preferences, thereby generating a self-organized path system with minimal detours. Through microsimulation and self-organization theory, the model successfully reproduces complex spatiotemporal patterns such as unidirectional pedestrian flows, reveals a strong dependence of walking efficiency on architectural geometry, and demonstrates that moderately reducing walkable area can enhance overall pedestrian throughput.

This study addresses the dual challenges of collision risk and traffic efficiency arising from pedestrian–vehicle interactions in urban environments by developing a two-dimensional grid-based multi-agent simulation framework that explicitly models streets, sidewalks, crosswalks, and obstacles. The authors introduce a weight-adjustable A* algorithm to parameterize diverse behavioral patterns—ranging from cautious to reckless—for both pedestrians and vehicles, enabling dynamic path planning and realistic modeling of complex traffic interactions. Experimental results demonstrate that obstacle density, traffic control infrastructure, and behavioral deviations significantly influence both collision likelihood and throughput efficiency. The proposed approach exhibits strong adaptability and effectiveness across multiple scenarios, offering a scalable and extensible simulation platform for safety evaluation in intelligent transportation systems.

This study addresses a critical gap in robotic evacuation strategies, which typically prioritize obstacle avoidance and macroscopic crowd flow while neglecting environmental affordances and human spatial expectations, thereby compromising both passive safety and psychological comfort. Through virtual reality evacuation experiments integrating behavioral psychology assessments and four multi-robot yielding strategies—Hide, LineEscape, Freeze, and ShortestPath—the research reveals for the first time the pivotal role of environmental semantics (e.g., refuge alcoves) in shaping human cognitive expectations. Findings demonstrate that actively leveraging spatial structures significantly enhances psychological comfort, that violations of expectation incur measurable cognitive costs, and that strategy preference follows Hide > LineEscape > Freeze > ShortestPath, confirming that proactive yielding outperforms freezing or efficiency-driven approaches. Moreover, prior human–robot interaction experience facilitates comprehension of complex social intentions.

This study addresses the limitations of existing evacuation models, which often assume fully rational agents with perfect global knowledge and thus fail to capture the heterogeneity and complexity of human behavior under emergency conditions. To overcome this, the authors propose a unified agent-based modeling framework that integrates cognition, emotion, social interaction, and personality traits. A key innovation is the explicit incorporation of neuroticism into a continuous fear dynamics model, combined with event certainty, memory-driven exit knowledge, individualized decision thresholds, and a forgetting mechanism. Simulation results demonstrate that the model effectively reproduces empirically observed phenomena such as evacuation delays, crowd confusion, injuries, and social influence effects, revealing how cognitive constraints, emotional fluctuations, and personality differences significantly impair evacuation efficiency.

Traditional statistical physics models struggle to capture agents with forward-looking behavior—such as pedestrians—because they rely solely on current or past states. This work proposes a novel statistical physics framework that models agent decisions as responses to anticipated future states. By introducing an observation-based cost function, the approach maps d-dimensional forward-looking agents onto (d+1)-dimensional non-anticipatory chains, enabling analysis through polymer physics methods. The study is the first to systematically incorporate foresight into statistical physics, introducing the concept of an “anticipation horizon” that naturally unifies operational and tactical modeling layers. Remarkably, even with the simplest cost function, the model successfully reproduces complex scenarios such as navigating crowded spaces and alighting from trains, outperforming existing state-of-the-art approaches.