Abstract
The future of transportation will be autonomous vehicles, which communicate with each other making smart and intelligent decisions. For example, vehicles need not to stop at intersections when vehicles autonomously coordinate themselves for the order of crossing. Cooperative decision-making has the potential to solve challenging traffic management problems and enhance transportation safety and efficiency. Nevertheless, the ideal communication and coordination protocols for the Connected and Autonomous Vehicles (CAVs) have unexpected security concerns. Self-interested vehicles may not always want to cooperate. We consider an advanced CAV network in which vehicles can directly communicate with each other sharing intentions and other information such as location and speed. Game theory is applied to study the interactions of CAVs in a conflicting environment. Both cooperative and noncooperative scenarios are considered, especially when one party may be untruthful (i.e., lying to gain advantage, e.g., crossing intersection first while asking other vehicles to slow down). The untruthful player benefits at the cost of the cooperative players. Socially optimal game outcomes are only possible when players are cooperative. Through game theoretical study, we identify two preventive measures, i.e., speed limits and safety gaps, which may be dynamically adjusted to induce CAVs to play truthfully thus reaching the socially optimal solution.