Dynamic congestion pricing based on user-optimal stochastic learning models

This paper describes the development and implementation of a dynamic congestion pricing scheme, augmented with a game-theoretic evolutionary learning model, into a realistic urban highway in Athens, Greece. The model recognizes several important behavioral features related to the response of users to the congestion pricing strategy. Such features include day-to-day as well as within-day adjustment mechanisms in the departure time and route choice behavior of users with respect to changes in travel cost, and heterogeneous classes of road users. The results can provide useful insights concerning the implications of modeling the stochastic learning processes of users for the efficient deployment of dynamic congestion pricing schemes in urban highways. The study findings can facilitate the evaluation of the acceptability of alternative congestion pricing schemes by users and the revenue management of system operators.