Bi-Level Programming Model for Dynamic Reversible Lane Assignment

To improve the space utilization and traffic capacity of a signalized junction, this paper proposes a dynamic reversible lane assignment method for approaches of signalized junctions that consider the game equilibrium between road users and traffic controllers. To theoretically analyze the behaviors of the players involved in the leader-follower strategic game, a bi-level programming model is established. To minimize the total queue length of approaches at signalized junctions, the upper model dynamically optimizes the reversible lane assignments and can be solved with the enumeration method or the Monte Carlo algorithm. The lower model outputs the traffic assignment at road sections using a Logit-based stochastic user equilibrium model that is solved by the method of successive averages. The general impedance of road section in lower model consists of the road travel time cost and the intra-time cost at the signalized junction connected with current road section. In addition, the interaction between the two levels is simulated in an iterative optimization procedure. Finally, this paper uses two numerical experiments to validate the proposed approach.