Evolution of Driver Behavior in the Presence of Automated Vehicles

While Automated Vehicles (AVs) are predicted to integrate into public roads sooner than expected, a completely connected, automated driving environment with no human-driven vehicles is yet to come. In light of this, a growing number of studies have focused on evaluating AVs' impact on the safety and efficiency of the transportation systems in a mixed traffic environment with both automated and traditional vehicles. However, the main assumption in most of these studies is that humans' behavior remains the same under repeated cycles of human-AV interactions over time. Adopting the theory of evolution in game settings, the present study develops a simulation framework to investigate potential changes in humans' behavior with the advent of automated vehicles. A mixed traffic environment is simulated where agents randomly interact with each other and play Stag-Hunt games while keeping track of their observations over time. AVs are assumed to be cooperative actors, while humans can either act individually and guarantee a predefined amount of payoff or take some risks and cooperate with their partner seeking higher benefits. Simulation results indicate that beyond social norms, humans' behavior depends on a measure of trust and what they believe about other actors in the system. If AVs' driving behavior offers sensible benefits over that of humans, repeated cycles of human-AV interactions over time can redefine trust levels and lower the associated risks of behavioral changes for humans. As individual beliefs change, humans may start to show different behaviors as well. It is thus crucial to account for the potential changes in humans' behavior when evaluating the overall impact of AVs on traffic flow dynamics.