Autonomous Driving at Unsignalized Intersections: A Review of Decision-Making Challenges and Reinforcement Learning-Based Solutions

Autonomous driving at unsignalized intersections is still considered a challenging application for machine learning due to the complications associated with handling complex multi-agent scenarios characterized by a high degree of uncertainty. Automating the decision-making process in these safety-critical environments involves comprehending multiple levels of abstraction associated with learning robust driving behaviors to enable the vehicle to navigate efficiently. In this survey, we aim at exploring the state-of-the-art techniques implemented for decision-making applications, with a focus on algorithms that combine Reinforcement Learning (RL) and deep learning for learning traversing policies at unsignalized intersections. The reviewed schemes vary in the proposed driving scenario, in the assumptions made for the used intersection model, in the tackled challenges, and in the learning algorithms that are used. We have presented comparisons for these techniques to highlight their limitations and strengths. Based on our in-depth investigation, it can be discerned that a robust decision-making scheme for navigating real-world unsignalized intersection has yet to be developed. Along with our analysis and discussion, we recommend potential research directions encouraging the interested players to tackle the highlighted challenges. By adhering to our recommendations, decision-making architectures that are both non-overcautious and safe, yet feasible, can be trained and validated in real-world unsignalized intersections environments. Note to Practitioners—Navigating unsignalized intersections is one of the most challenging aspects of urban autonomous driving, directly linked to high accident rates and inefficiencies in road traffic. This paper surveys decision-making strategies for autonomous vehicles at such intersections, aiming to bridge the gap between theoretical advancements and real-world applications. Traditional rule-based decision-making and optimization techniques can be effective in predictable scenarios but often fail in environments with high uncertainty and dynamic multi-agent interactions. Reinforcement learning (RL)-based methods, while promising in their ability to learn from experience and adapt to complex environments, still face significant challenges. These include ensuring safety-critical behavior, achieving real-time performance, and bridging the simulation-to-reality gap. For practitioners in the autonomous driving and robotics industries, this survey provides a practical guide to evaluating existing decision-making approaches, their potential for deployment, and their limitations. It highlights key considerations such as computational efficiency, scalability, and robustness required for urban driving scenarios. The discussion also emphasizes the importance of integrating diverse data sources and testing solutions in real-world environments to ensure reliability. By leveraging the insights presented in this survey, practitioners can better assess and refine algorithms to tackle the nuanced challenges of unsignalized intersections. This work serves as a foundational step towards developing more robust, adaptable, and safe decision-making systems for autonomous vehicles in such complex urban environments.