Fairness concern-based coordinated vehicle route guidance using an asymmetrical congestion game

In this paper, an asymmetrical congestion game is used to build a fairness concern-based coordinated route guidance model for alleviating traffic congestion. Coordinated vehicle route guidance is commonly recognized as an effective way to alleviate the "route flapping" phenomenon, where a new congestion appears since numerous vehicles obey the same guidance in selfish route diversion. The existing studies mainly use the classic (symmetrical) congestion games, whose payoff depends only on the number of chosen players, to calculate the flow proportion for the alternative roads. This mechanism may make some front vehicles receive the worse payoffs than the rear ones, which further results in the front vehicles being unwilling to execute the recommended paths due to their unfair psychology. Players' fairness concern in game theory has been noted in some fields (such as supply chain) but not traffic route guidance. In this work, to fill the above gap, the concept of vehicular fairness concern is introduced and the corresponding mathematical expression is formulated. An asymmetrical congestion game-based coordination route model is built, which can improve the fairness coefficient of vehicles. Moreover, a reinforcement learning-based approximated solving algorithm is designed since the existing solving theorems are not necessarily valid for the asymmetrical congestion game. An experimental study shows that (1) the Nash equilibrium coefficients generated by the approximated solving algorithm all converge to 1.00, and (2) the averaged fairness coefficient is increased from 0.92 to 0.96 on the basis of equal travel time due to the improved payoff functions.