A Visual Analytics Approach to Traffic Signal Control Optimization for Multiple Road Intersections

Traffic lights have been widely used to alleviate traffic congestion and improve traffic efficiency. However, due to the complex variability of traffic conditions and interactions between intersections, formulating ef-fective traffic signal control strategies for multiple intersections is highly challenging. Aiming at issues in existing optimization methods which are adaptive and fixed-time, this paper proposes a visual analytics method for multi-intersection traffic signal control optimization. Through collaborative interviews with experts, the paper extracts the problems and demands related to identifying and optimizing inefficient traf-fic signal control strategies, and accordingly constructs a novel visual analytics system—VisualTSC. The implemented trajectory mining process, which integrates the extraction of inefficient intersections and generation of optimization strategies, assists experts to understand the traffic efficiency of intersections and provides feasible alternative control strategies via simulation combined with the visual analytics sys-tem. VisualTSC employs a multi-level contour map to visualize intersection relationships, and offers tabu-lar and trajectory views to support optimal strategies decision-making. The effectiveness of VisualTSC is verified via two application scenarios using real-world datasets and follow-up expert interviews, where experts reproduced the cases and provided positive qualitative evaluations on its performance.