A Game Theory Model for Self-adapting Traffic Flows with Autonomous Navigation

It is widely believed that road traffic as a whole self-adapts to the current situation to make travel times shorter, if the navigation devices exploit real-time traffic information. A novel theoretical approach to study this belief is the online routing game model. This chapter describes the model of online routing games in order to be able to determine how we can measure and prove the benefits of online real-time data in navigation systems. Three different notions of the benefit of online data and two classes of online routing games are defined. The class of simple naive online routing games represents the current commercial car navigation systems. Simple naive online routing games may have undesirable properties: stability is not guaranteed, single flow intensification may be possible and the worst case benefit of online data may be bigger than one, i.e. it may be a “price”. One of the approaches to avoid such problems of car navigation is intention propagation where agents share their intention and can forecast future travel times. The class of simple naive intention propagation online routing games represents the navigation systems that use shortest path planning based on forecast future travel times. In spite of exploiting intention propagation in online routing games, single flow intensification may be possible, the traffic may fluctuate and the worst case benefit may be bigger than one. These theoretical investigations point out issues that need to be solved by future research on decision strategies for self-adapting traffic flows with autonomous navigation.