Behavior-centric collision avoidance for realistic crowd simulation

Abstract

Crowd simulation, which is widely used for computer animations and safety engineering, is the pro-cess of giving the appearance or effect of interaction amongst a large number of people together synthetically. A Crowd simulation model falls into one of three categories: social force models (SFM), cellular automata models (CAM), and agent-based models (ABM). In spite of the many technical progresses in each category, the collision avoidance steering behavior of a virtual crowd in a simulation is still unrealistic. This dissertation proposes a collision avoidance behavior model based on psychological findings of human behaviors such as gaze-movement angle (GMA), side stepping, gait motion, and personal reaction bubble (PRB) to generate a realistic collision avoidance steering behavior of a virtual crowd. By calculating the GMA between agents, collision can be predicted and avoided without knowing the exact trajectories of the agents. PRB guarantees realistic space-keeping between agents. The proposed steering model consists of four phases: (1) GMA-based collision prediction for mid/long range by using speed-variant Information Process Space (IPS), (2) collision avoidance steering, (3) gait-based locomotion generation, and (4) space-keeping based on PRB. The effectiveness of the proposed speed-variant IPS was tested on various types of agent flows with different densities. The total loss of kinetic energy accumulated during an agent’s movement and the ratio of the length of the path actually traveled to the length of the original path are used as key metrics to figure out the features between the different types of flows. Finally, to show the validity of the proposed model, the comparison is carried out with well-known fundamental diagrams.