Abstract:For walking pedestrians, when they are blocked by obstacles or other pedestrians, they adjust their speeds and directions to avoid colliding with them, which is called collision avoidance behavior. This behavior is the most complex part of pedestrians' walking processes and its modeling and simulation are the keys to realistic crowd simulation, which serves as the foundation for various applications. However, most existing methods either lack the representation power to accurately model the complex collision behavior or do not model it explicitly, which leads to a poor level of realism of the simulation. To realize realistic crowd simulation, we propose to analyze, understand, and model the collision avoidance behavior in a data-driven way. First, to automatically detect collision avoidance behavior for further analysis, we propose a domain transformation algorithm that detects it by transforming the trajectories in the spatial domain into a new domain where the behavior is much more apparent and is thus easier to detect. The new domain also provides a new perspective for understanding collision avoidance behavior. Second, since there are no mature metrics to evaluate the level of realism, we propose a new evaluation metric based on the least-effort theory, which evaluates the realism of collision avoidance behavior by its physical and mental consumption. This evaluation metric also provides the foundation of modeling. Third, for realistic crowd simulation, we design a reinforcement learning model. It trains agents with our proposed reward function that models pedestrians' intrinsic needs of "reducing effort consumption'' and thus can guide agents to behave realistically when avoiding collisions. Extensive experiments show our model is 55.9% and 52.5% more realistic in collision avoidance behavior than the best baselines on two real-world datasets. We release our codes at https://github.com/tsinghua-fib-lab/TECRL.

Experimental study on pedestrians' collision avoidance

Pedestrians’ behavior in emergency evacuation: Modeling and simulation*

Collision Avoidance Mechanism for Pedestrian Interactions

A modified social force model considering collision avoidance based on empirical studies

Experimental Study on the Interaction Mechanism of Cross-Walking Pedestrians

A Modified Heuristics-Based Model for Simulating Realistic Pedestrian Movement Behavior

Understanding and Modeling Collision Avoidance Behavior for Realistic Crowd Simulation

Experimental study on the movement characteristics of pedestrians under sudden contact forces

Modeling pedestrian flow accounting for collision avoidance during evacuation

Experimental Study on Unidirectional Pedestrian Descending and Ascending Stair With a Fixed Obstacle

Deconstructing Pedestrian Crossing Decisions in Interactions with Continuous Traffic: an Anthropomorphic Model

Experimental Study on Movement Characteristics of Pedestrians with Different Speeds

Pedestrian Evacuation Modelling with Dynamics Congestion Avoidance

Deconstructing Pedestrian Crossing Decision-making in Interactions with Continuous Traffic: an Anthropomorphic Model

A Modified Social Force Model Considering Relative Velocity of Pedestrians

A speed-based model for crowd simulation considering walking preferences

Improved Social Force Model based on Pedestrian Collision Avoidance Behavior in Counterflow

Experimental Study of High-Density Pedestrian Flow Field Characteristics at a Crossing

Kinetic and Kinematic Features of Pedestrian Avoidance Behavior in Motor Vehicle Conflicts

Scene-Based Pedestrian Safety Performance Model in Mixed Traffic Situation

Experimental Investigation on the Pair Interaction Between Pedestrians