This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
This paper focuses on learning a general-purpose policy that considers the heterogeneity of selfish intersections and the influence of neighboring intersections in urban-scale traffic signal control (TSC). To overcome the limitations of existing methods that only consider information about selfish intersections, we present the CityLight model, which includes a Neighbor Influence Encoder, which explicitly models the influence of neighboring intersections, and a Neighbor Influence Aggregator, which aggregates influences by considering the competitive relationships between neighboring intersections. CityLight demonstrates its effectiveness on five city datasets of varying scale, demonstrating an average throughput improvement of 11.68% and a generalization performance improvement of 22.59%.
Takeaways, Limitations
•
Takeaways:
◦
We present a novel method to improve city-scale traffic signal control performance by considering the influence of neighboring intersections as well as selfish intersections.
◦
We demonstrate the practicality of our universal policy by validating its effectiveness on urban datasets of various sizes.
◦
Suggesting the possibility of more efficient traffic signal control by considering the competition between neighboring intersections.
•
Limitations:
◦
Further analysis of the computational cost and scalability of the proposed model is needed.
◦
Further research is needed on the model's robustness to various traffic conditions (e.g., accidents, special events).
◦
Additional experiments and validation are needed for application to real-world urban environments.
