COLLISION AVOIDANCE APPROACH FOR EXAMPLE-BASED CROWD SIMULATION

Authored by Weiwei Xing, Lili Zhu, Xiang Wei, Peng Bao

Date Published: 2018

DOI: 10.24507/ijicic.14.01.127

Sponsors: Chinese National Natural Science Foundation

Platforms: No platforms listed

Model Documentation: Other Narrative Pseudocode Mathematical description

Model Code URLs: Model code not found

Abstract

Collision avoidance plays an essential role in crowd simulation and determines the simulation quality to some extent. We propose a novel method to improve the collision avoidance performance based on the existing crowd simulation methods. The example-based method is chosen as the basic method by reasons of the realistic simulation result. Considering collision avoidance in the example-based method relies heavily on data captured from the real world, velocity obstacle (VO) is introduced to calculate an action of an agent when collision-free example cannot be found. We finally optimize the performance of collision avoidance by performing the novel collision fixing (CF) algorithm. The CF takes advantage of obtained actions to detect potential collisions and employs repulsive force to avoid the collisions. Experiments are conducted to evaluate the performance of the proposed method, and the results show that the method greatly reduces collision times, especially in complex simulation scenarios, while keeping the scenarios otherwise realistic.

Tags

Agent-based modeling Agents Crowd simulation Model Collision avoidance Example-based simulation