Simulação de agentes virtuais dotados de percepção sintética em situações de emergência

Abstract

This work presents a model to provide perceptual abilities (vision, smelling and hearing) to virtual agents, focusing on applications of crowd simulation for emergency situations. The models for synthetic vision and hearing were based on papers published by known authors, while the virtual hearing was modeled in details. The synthetic hearing model was based on real characteristics, such as sound attenuation caused by walls and distance from the sound source. Also, some physiological behaviors of the human hear were considered, such as the natural diffculty of a human being to hear some frequency bands, and hearing losses due to aging. Finally, the hearing model also included masking effects, in which some sounds may be masked by others with similar frequencies.

The proposed perceptual model was evaluated in applications of crowd simulation for emergency situations, in which an alarm was placed into different positions of the virtual environment, and the agents had to decide what to do based on their percept