Exploring the impact of traffic signal control and connected and automated vehicles on intersections safety: A deep reinforcement learning approach

The document explores the impact of combining connected and automated vehicles (CAVs) with deep reinforcement learning (DRL) traffic signal control methods on intersection safety, providing insights into the potential reduction of conflicts and offering recommendations for future research and policy considerations.

Abstract

In transportation networks, intersections pose significant risks of collisions due to conflicting movements of vehicles approaching from different directions. To address this issue, various tools can exert influence on traffic safety both directly and indirectly. This study focuses on investigating the impact of adaptive signal control and connected and automated vehicles (CAVs) on intersection safety using a deep reinforcement learning approach. The objective is to assess the individual and combined effects of CAVs and adaptive traffic signal control on traffic safety, considering rear-end and crossing conflicts. The study employs a Deep Q Network (DQN) to regulate traffic signals and driving behaviors of both CAVs and Human Drive Vehicles (HDVs), and uses Time To Collision (TTC) metric to evaluate safety. The findings demonstrate a significant reduction in rear-end and crossing conflicts through the combined implementation of CAVs and DQNs-based traffic signal control. Additionally, the long-term positive effects of CAVs on safety are similar to the short-term effects of combined CAVs and DQNs-based traffic signal control. Overall, the study emphasizes the potential benefits of integrating CAVs and adaptive traffic signal control approaches in order to enhance traffic safety. The findings of this study could provide valuable insights for city officials and transportation authorities in developing effective strategies to improve safety at signalized intersections.

Method

The method of the document involves exploring the impact of traffic signal control and connected and automated vehicles (CAVs) on intersection safety using a deep reinforcement learning (DRL) approach. The study includes a review of existing literature, a detailed methodology for the study, which encompasses the DRL approach to control traffic signals, car-following models to represent driving behavior, safety metrics, and the definition of different types of conflicts. The research also involves simulations to demonstrate the impact of DRL-based traffic signal control and CAVs on safety at intersections. Additionally, the document provides recommendations for future work, such as exploring alternative types of deep Q networks (DQNs) and advanced reinforcement learning techniques to further enhance traffic signal control.

Main Finding

The main finding of the document is that the combined implementation of connected and automated vehicles (CAVs) and deep reinforcement learning (DRL) based traffic signal control, using a Deep Q Network (DQN), significantly reduces rear-end and crossing conflicts at intersections. Additionally, the study highlights that the long-term positive effects of CAVs on safety are similar to the short-term effects of combined CAVs and DRL-based traffic signal control, emphasizing the potential benefits of integrating CAVs and adaptive traffic signal control approaches to enhance traffic safety.

Conclusion

The conclusion of the document highlights the significant impact of integrating connected and automated vehicles (CAVs) with a Deep Q Network (DQN)-based traffic signal control approach on traffic safety at signalized intersections. The study demonstrates that both CAVs and DQN-based traffic signal control independently contribute to improving safety at intersections, and when combined, their effect is even more substantial in achieving collision-free intersections. Furthermore, the research suggests that the impact of CAVs and DQN-based traffic signal control remains similar regardless of the penetration rate of CAVs in fixed-time traffic signal scenarios, indicating comparable short-term effects to the long-term effects of CAVs alone on traffic safety. The document also provides recommendations for future research, such as exploring alternative types of DQNs, investigating advanced reinforcement learning techniques, and expanding the study to multiple intersections or a broader network to further enhance traffic signal control and traffic safety.

Keywords

Deep Q learning, Connected and Automated Vehicles, Traffic Safety, Intersection Safety, Adaptive Signal Control.