Intuitive and Privacy-Preserving Traffic Light Control System for Autonomous Vehicles

An efficient traffic light control system (TLCS) is an integral part of monitoring the flow of autonomous vehicles (AVs) through road junctions. Existing TLCS systems have limitations, focusing on specific scenarios like emergency vehicles or considering only the queue length at intersections. Furthermore, the absence of privacy protection in these systems exposes vehicles to potential tracking risks. We propose an intuitive and privacy-preserving TLCS (IPTLCS) to improve the performance of the TLCS for multiple traffic scenarios and prevent tracking of vehicles at traffic signals. The proposed IPTLCS uses a deep Q-learning (DQN) algorithm based on the quantity versus priority concept, adaptive to all scenarios, including pedestrians and emergency vehicles, making the framework applicable to real-time situations. Further, we propose a novel anonymity preserving protocol (APP) to protect the privacy of vehicles using two-party computation (2PC) that can prevent the tracking of AVs in our environment. Extensive experimental studies of the IPTLCS reveal that the model can achieve a reduced waiting time of 3.45 s/vehicle, a queue length of 3.32 vehicles/lane, a run time of 0.002 s, and a communication overhead of 3 KB. The increase in the efficiency of the proposed IPTLCS model in comparison with existing models in terms of waiting time, queue length, run time, and communication overhead is 11.08%–43.25%, 2.35%–24.23%, 33.3%–65.5%, and 38.77%–54.38%, respectively. Impact Statement The infusion of deep learning (DL) methodologies into traffic light systems signifies a substantial leap in advancing the management of AVs on roadways. Confronting the limitations of current traffic light systems, especially in adapting to diverse scenarios with varying quantities and priorities of vehicles, this study introduces a framework for efficient AV navigation that minimizes traffic collisions with reduced delays. Incorporating the proposed privacy measures ensures vehicular data protection, enhancing the overall system’s safety. Results demonstrate that the proposed method addresses the secure management of vehicular data and achieves reduced processing times, paving the way for progressing more intelligent and secure urban mobility solutions and fostering a seamless coexistence between AVs and pedestrians.