CTRL: Cooperative Traffic Tolling via Reinforcement Learning

Hexi Jin, Yiheng Wang, Guanjie Zheng

Abstract:

People have been working long to tackle the traffic congestionproblem. Among the different measures, traffic tolling has beenrecognized as an effective way to mitigate citywide congestion.However, traditional tolling methods can not deal with the dynamictraffic flow in cities. Meanwhile, thanks to the development oftraffic sensing technology, how to set appropriate dynamic tollingaccording to real-time traffic observations has attracted researchattention in recent years.

In this paper, we put the dynamic tolling problem in a reinforce-ment learning setting and try to tackle the three key challengesof complex state representation, pricing action credit assignment,and route price relative competition. We propose a soft actor-criticmethod with (1) a route-level state attention, (2) an interpretable andprovable reward design, and (3) a competition-aware Q attention.Extensive experiments on real datasets have shown the superiorperformance of our proposed method. In addition, interesting anal-ysis on pricing actions and vehicle routes have demonstrated whythe proposed method can outperform baselines