Real-time crowd simulation in virtual urban environments using adaptive grids

Crowd simulation is a relatively new research area, attracting increasing attention from both academia and industry. This thesis proposes Adaptive Grids, a novel hybrid approach for controlling the behavior of agents in a virtual crowd. In this approach, the motion of each agent within the crowd is planned considering both global and local path planning strategies. For global path planning, a cellular adaptive grid is constructed from a regular navigation map that represents the 2-D topology of the simulation terrain. A navigation graph with efficient size is then pre-computed from the adaptive grid for each possible agent goal. Finally, the navigation graph is used to generate a potential field on the adaptive grid by using the connectivity information of the irregular cells. Global path planning per agent has constant time complexity. For local path planning, Helbing Traffic-Flow model is used to avoid obstacles and agents. Potential forces are then applied on each agent considering the local and global decisions of the agent, while providing each agent the freedom to act independently.