Lane-changing strategies of autonomous vehicles and social dilemmas in mixed traffic: A simulation study

This study investigates the impact of different lane-changing strategies of autonomous vehicles (AVs) on traffic dynamics and social efficiency in mixed traffic conditions. We introduce a multi-agent traffic model based on a cellular automaton framework, incorporating human-driven vehicles (HDVs) and three types of AVs: non-lane-changing (AV), cooperative (AV-C), and permissive (AV-D). Each AV type follows distinct longitudinal and lateral rules under Adaptive Cruise Control (ACC) or Cooperative ACC (CACC). The simulation results reveal that non-lane-changing AVs maximize traffic flow but struggle with obstacle avoidance. AV-C agents maintain platoon integrity, while AV-D agents improve maneuverability at the cost of platoon stability. We analyze the emergence of social dilemmas using the Social Efficiency Deficit (SED) metric and identify conditions under which individual rationality conflicts with global traffic performance. The findings highlight the need for hybrid control strategies and external incentives to support early-stage AV deployment and ensure cooperative equilibria.