Evaluation and Optimization Method for Urban Signal Control Performance Based on Automatic Vehicle Identification Data
This paper proposes a closed-loop methodology that leverages Automatic Vehicle Identification (AVI) data to infer signal control logs, perform fine-grained cycle-level performance evaluation, and optimize timing via a Plant Root Growth Optimization (PRGO) algorithm, effectively reducing vehicle delays and enabling a transition from experience-based to data-driven urban signal control.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Traffic lights are the silent conductors of our daily commutes, yet for city planners, they often operate in the dark. While modern sensors can track every car moving through an intersection, the actual instructions given to those lights—the precise moments they turn green or red—remain hidden inside the controllers themselves. This creates a frustrating blind spot: engineers have a flood of data about where cars are, but they lack the blueprint of how the lights are actually behaving. Without knowing the exact timing sequence, it is difficult to tell if a traffic jam is caused by too many cars or by a poorly timed signal. This gap between what is visible on the road and what is hidden in the machine has long prevented cities from moving beyond guesswork toward truly data-driven traffic management.
A team of researchers from Shandong Jiaotong University in China has developed a way to peer behind that curtain using only the data from the cars themselves. By analyzing the movement patterns of vehicles equipped with automatic identification systems, they created a method to reverse-engineer the signal control logic. Instead of needing direct access to the traffic light's internal computer, their system listens to the rhythm of the traffic. It identifies when cars bunch up and when they flow freely, using these patterns to deduce exactly when the light turned green, how long the cycle lasted, and which lanes were allowed to move together. This process effectively turns the invisible timing plan into a visible, measurable reality, allowing for a detailed inspection of how well the intersection is performing.
The researchers tested this approach on a busy intersection in Jinan, comparing their inferred data against the actual, hidden signal logs to see how accurate their method was. The results were remarkably precise. The system estimated the length of the signal cycle with an error of less than five seconds and pinpointed the exact moment a light turned green with an error of just two seconds. With this accurate timeline in hand, the team could then calculate specific performance metrics that were previously impossible to measure without the internal data. They determined how full the lanes were, how long the lines of waiting cars grew, and exactly how much time drivers were losing to delays. They found that while some directions flowed smoothly, others were severely congested, with certain lanes operating at a level of saturation that indicated significant inefficiency.
To make sense of this complex data, the researchers used a clustering technique to group the intersection's performance into distinct levels, ranging from free-flowing to heavily congested. This analysis revealed that the westbound through lane was the primary bottleneck, operating in a poor state for nearly half of the morning rush hour. Armed with this diagnosis, the team built a new optimization model designed to minimize the total time vehicles spent waiting. They tested three different mathematical strategies to find the best solution, including two well-known methods and a newer approach inspired by how plant roots grow to find nutrients in the soil. This root-growth method, which mimics the way roots explore and branch out to locate resources, proved to be the most effective at navigating the complex search for the perfect timing plan.
When the researchers applied the new timing plan generated by this root-growth method to a computer simulation of the intersection, the improvements were clear. The average delay for vehicles dropped by nearly 9.34 percent, and the maximum length of the traffic queues decreased significantly. The westbound lane, which had been the worst performer, saw a delay reduction of over 18 percent, proving that the system could target specific problem areas effectively. While these results come from a simulation rather than a live street test, the study demonstrates a viable path forward for cities. It shows that by listening closely to the data already being generated by vehicles, traffic engineers can uncover the hidden logic of their signals, diagnose specific failures, and adjust the lights to keep the city moving more efficiently, all without needing to install new hardware or access proprietary controller systems.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.