Cooperative and Interaction-aware Driver Model for Lane Change Maneuver (2403.01752v1)

Abstract: To achieve complete autonomous vehicles, it is crucial for autonomous vehicles to communicate and interact with their surrounding vehicles. Especially, since the lane change scenarios do not have traffic signals and traffic rules, the interactions between vehicles need to be considered for the autonomous vehicles. To address this issue, we propose a cooperative and interaction-aware decision-making algorithm for autonomous vehicles that stochastically considers the future behavior of surrounding vehicles based on actual driving data. The algorithm is designed for both lane changing and lane keeping vehicles, and effectively considers interaction by using an interaction model based on relative information between vehicles with fewer states. To design the decision-making, the interaction model is defined as Markov decision process, and stochastic dynamic programming is used to solve the Markov decision process. We validate the effectiveness of our proposed algorithm in lane change scenarios that require interaction. Our results demonstrate that the proposed algorithm enables cooperative and interaction-aware decision-making while accommodating various driving styles. Additionally, by comparing it with other methods, such as the intelligent driver model and game theory-based decision-making, we validate the safety and comfortable decision-making of our proposed algorithm. Furthermore, through driving with a human-driven vehicle, it is confirmed that the proposed decision-making enables to cooperatively and effectively drive with humans.