A probabilistic agent approach to the trafic jam problem

Abstract

Artificial life simulations of social situations are a relative new field which aims to model situations which are too complex to be analytically investigated. In this paper, we develop commuter-agents with simple probabilistic models of the world and show that such agents can develop cooperation which aids the society as a whole. We show that there are situations in which the more powerful agents are sometimes forced by their greater knowledge into taking a lower utility than the weaker ones. In the last series of experiments we show that agents which have the ability to predict others' road usage can materially improve the utility of the population as a whole.