A Study in Human-Machine Interaction through Agent Simulation: an Application in Pedestrian Crossing

The goal of this research is to use agent simulation in realizing human-machine interaction, so that we can perform different aspects of experiments. The domain that we have chosen is the pedestrian-vehicle interaction in street crossing. The participants of the simulation include pedestrians, vehicles, and traffic lights, which are all represented by agents. The autonomous nature along with individual traits of agents enable the simulator to show the dynamic interaction among the agents. The attributes of agents like age, gender, and aggressiveness are considered along with different reaction time. A prototype system uses Jade as the platform in implementation. The experiments show a variety of results based on interaction among various agents.