Advanced models for centroidal particle dynamics: short-range collision avoidance in dense crowds.

Area of Science:

• Computational Social Science
• Agent-Based Modeling
• Physics of Complex Systems

Background:

• Computer simulations of dense crowds are crucial for event planning and safety analysis.
• Existing particle-based methods often unrealistically assume collision-free trajectories, failing to capture dense crowd dynamics.

Purpose of the Study:

• To introduce Centroidal Particle Dynamics (CPD), an agent-based method for simulating dense crowds.
• To improve the realism of crowd simulations by explicitly modeling personal space and collision avoidance.

Main Methods:

• Developed an agent-based model, Centroidal Particle Dynamics (CPD), that incorporates compressible personal space for each entity.
• Simulated dense crowd scenarios, including pedestrian distraction, flocking, and interactions with emergency vehicles.

Main Results:

• CPD reproduces key emergent dense crowd phenomena at the microscopic level with higher accuracy than state-of-the-art methods.
• Simulations show more visually convincing collision-avoidance paths and congruence with real trajectory data.
• Advanced models demonstrate emergent behaviors similar to real-world crowd observations.

Conclusions:

• CPD offers a more realistic approach to crowd simulation, capturing complex emergent behaviors.
• The method shows potential for safety-critical applications such as urban design and evacuation analysis.
• Further development can enhance confidence in CPD for real-world crowd-safety planning.