Travel Time Perception and Learning Mechanisms in Traffic Networks

Abstract

Models of travel time perception and learning mechanisms in traffic networks are presented. Mechanisms for updating travel times in light of new experiences and for triggering and terminating the updating process are examined. Travel time perception and learning are modeled on the basis of concepts from Bayesian statistical inference. Mechanisms for triggering and terminating the learning process are modeled with the use of simple heuristic rules based on the inter-update period and thresholds that define the salience of travel times and acceptable confidence levels. These models are embedded inside a microscopic (agent-based) simulation framework and model to study their collective effects on the day-to-day behavior of traffic flows through a series of experiments. The effect of alternative travel time perception and learning mechanisms on system performance and its dynamic properties, in particular convergence, is explored.