Multi-agent Simulation of Unorganized Traffic

Kamal Karlapalem
International Institute of Information Technology

Traffic simulation is one of the most complex simulation projects that can be undertaken. The main issues are: modeling of autonomous behavior of drivers, modeling of their interaction, and ability to simulate the traffic and procure reliable realistic results. Organized traffic with drivers heeding to well defined traffic rules is less dynamic and erratic, than modeling unorganized traffic, wherein the drivers either do not heed to well defined traffic rules, or there are no traffic rules in place. This paper shows the viability of applying multi-agent simulation for unorganized traffic. In particular, we model the behavior of drivers, as being cautious, normal, and aggressive, and show results about average speed of vehicles in traffic, number of overtakes, and number of accidents occurring with different proportions of aggressive and cautious drivers. A multi-agent simulator with graphics interface has been implemented to visualize and evaluate the traffic flow.