An Intelligent Decision-making Framework for Optimizing Left-Turn Restrictions: Reverse Engineering AI Predictions
Research Poster Engineering 2025 Graduate ExhibitionPresentation by Tanveer Ahmed
Exhibition Number 90
Abstract
Left-turn movements pose significant safety hazards and reduce the operational efficiency of signalized intersections. One effective strategy to mitigate these issues is to restrict conflicting left-turns at strategic locations. However, determining the optimal locations for such restrictions in large urban networks is challenging due to the complexity of traffic dynamics and the large solution space. AI models such as genetic algorithms can efficiently determine near-optimal solutions from large spaces. However, this study seeks to understand why these solutions are optimal. Therefore, a two-stage methodology is proposed where the first stage applies a bi-level optimization framework to determine near-optimal left-turn restriction locations. The second stage introduces a binary logit model that is used to explain the traffic parameters influencing these decisions. When applied to the Pittsburgh traffic network, the PBIL algorithm demonstrated up to a 12% reduction in travel time under peak demand without significantly increasing trip lengths. The logit model, trained on known demand scenarios, indicates that intersections with higher values of left-turning green-ratio, flow-ratio, and protected green-ratio are less likely to benefit from left-turn restrictions. Furthermore, the model's predictions for unknown demand levels can identify locations of left-turn restrictions using only information from individual intersections. Overall, the framework provides a data-driven approach for reverse engineering AI solutions and make them interpretable for implementation agencies.
Importance
Artificial Intelligence (AI) is revolutionizing problem-solving, but many AI models function as "black boxes," providing optimal solutions without clear explanations. This lack of interpretability is a challenge in transportation, where infrastructure and policy decisions involve significant capital investments. Relying solely on black-box AI is impractical for critical planning. This study aims to reverse-engineer an AI model to enhance interpretability, making solutions more transparent for transportation agencies. It focuses on optimizing left-turn restrictions in a large network, but the framework can be extended to other applications, such as identifying optimal infrastructure locations, determining project implementation sequences, or optimizing capital improvement plans and budgets. By improving AI transparency, this research supports informed decision-making in transportation infrastructure and policy planning.