Social-Psychology-Inspired Reinforcement Learning Framework for Conflict Management in Connected Vehicles

In any connected network, resource scarcity, perceived road blocks, and incongruent objectives can potentially ensue conflicts among stakeholders. In the existing literature, trust has been cited as a crucial component in effective conflict management (CM). Besides trust, empathy, and social intelligence (SI) play decisive roles in enhancing cooperation, encouraging information sharing, and promoting problem solving. In this article, we discuss the three major components of CM and propose a computational model, which is inspired from social psychology for CM in connected vehicles (CVs). Our mathematical algorithm focuses on three factors, namely, trust, empathy, and SI that are learned via social interactions among vehicles to ensure safety of vehicles and passengers. The triad of trust, empathy, and SI is used to aid reinforcement learning (RL) for obtaining the optimal
$q$
-values and rewards in the shortest duration of time in the CV network. We have examined how the three factors influence the learning process and analyzed their CM potentials. Results show that the proposed model is 118.18% more efficient than the trust-only-based RL algorithm.