Gestión inteligente de semáforos mediante visión y aprendizaje por refuerzo

• Caballero, Diego ^[1] ; Villa, Jaime ^[1] ; Fernández, Rubén ^[1] ; Abderrahim, Mohamed ^[1] ; Sanchis, Araceli ^[1] ; Armingol, José María ^[1]
  1. [1] Universidad Carlos III de Madrid

     Universidad Carlos III de Madrid

     Madrid, España

     
• Localización: Jornadas de Automática, ISSN-e 3045-4093, Nº. 47, 2026
• Idioma: español
• DOI: 10.17979/ja-cea.2026.47.13792
• Títulos paralelos:
  • Intelligent Traffic Signal Management through Vision and Reinforcement Learning
• Enlaces
  • Texto completo
• Resumen
  • español

    La congestión del tráfico urbano constituye uno de los principales retos de las ciudades modernas. Los sistemas tradicionales de control de semáforos, basados en tiempos fijos, no se adaptan a las variaciones dinámicas del tráfico. Este trabajo presenta un agente autónomo de control de semáforos basado en aprendizaje por refuerzo profundo (DRL) que utiliza únicamente imágenes fotorrealistas de la intersección como entrada, eliminando la dependencia de costosas infraestructuras de sensores. La validación se realiza en una plataforma de co-simulación que combina CARLA, para el realismo visual, y SUMO, para la dinámica microscópica del tráfico. Los resultados muestran reducciones del tiempo de espera de hasta el 22,68% en la intersección de entrenamiento, una mejora del 23,76% al reentrenar el agente en una intersección con topología distinta y una sinergia positiva en escenarios multi-agente, donde la cooperación entre intersecciones adyacentes mejora la recompensa hasta un 59,78%.

  • English

    Urban traffic congestion remains a major challenge in modern cities. Traditional fixed-time traffic signal controllers cannot adapt to dynamic traffic conditions. This work presents an autonomous traffic signal control agent based on Deep Reinforcement Learning (DRL) that uses only photorealistic images of the intersection as input, removing the dependency on costly physical sensor infrastructure. The validation is performed in a high-fidelity co-simulation platform combining CARLA, for visual realism, and SUMO, for microscopic traffic dynamics. Results show reductions of up to 22.68% in average waiting time at the training intersection, an improvement of 23.76% when retraining the agent in an intersection with a different topology, and positive synergy in multi-agent settings, where cooperation between adjacent intersections improves the reward by up to 59.78%.

• Referencias bibliográficas
  • Arulkumaran, K., Deisenroth, M. P., Brundage, M., Bharath, A. A., 2017. Deep reinforcement learning: A brief survey. IEEE Signal Processing...
  • Benhamza, K., Seridi, H., Agguini, M., Bentagine, A., 2024. A multi-agent reinforcement learning based approach for intelligent traffic signal...
  • Bouktif, S., Cheniki, A., Ouni, A., El-Sayed, H., 2023. Deep reinforcement learning for traffic signal control with consistent state and reward...
  • Cai, C., Wei, M., 2024. Adaptive urban traffic signal control based on enhanced deep reinforcement learning. Scientific Reports 14. DOI: 10.1038/s41598-024-64885-w
  • Dosovitskiy, A., Ros, G., Codevilla, F., Lopez, A., Koltun, V., 2017. CARLA: An open urban driving simulator. In: Proceedings of the 1st Annual...
  • Eom, M., Kim, B.-I., 2020. The traffic signal control problem for intersections: a review. European Transport Research Review 12, 1–20. DOI:...
  • Huber, P. J., 1964. Robust estimation of a location parameter. The Annals of Mathematical Statistics 35 (1), 73–101. DOI: 10.1214/aoms/1177703732
  • Kingma, D. P., Ba, J., 2014. Adam: A method for stochastic optimization. CoRR abs/1412.6980.
  • Lee, J., Chung, J., Sohn, K., 2020. Reinforcement learning for joint control of traffic signals in a transportation network. IEEE Transactions...
  • Li, D., Wu, J., Xu, M., Wang, Z., Hu, K., 2020. Adaptive traffic signal control model on intersections based on deep reinforcement learning....
  • Liang, X., Du, X., Wang, G., Han, Z., 2019. A deep reinforcement learning network for traffic light cycle control. IEEE Transactions on Vehicular...
  • Lieberthal, E. B., Serok, N., Duan, J., Zeng, G., Havlin, S., 2024. Addressing the urban congestion challenge based on traffic bottlenecks....
  • Lopez, P. A., Behrisch, M., Bieker-Walz, L., Erdmann, J., Flötteröd, Y.-P., Hilbrich, R., Lücken, L., Rummel, J., Wagner, P., Wiessner, E.,...
  • Mnih, V., Kavukcuoglu, K., Silver, D., Rusu, A. A., Veness, J., Bellemare, M. G., Graves, A., Riedmiller, M., Fidjeland, A. K., Ostrovski,...
  • Robertson, D. I., Bretherton, R. D., 1991. Optimizing networks of traffic signals in real time—the SCOOT method. IEEE Transactions on Vehicular...
  • Sims, A. G., Dobinson, K. W., 1980. The Sydney coordinated adaptive traffic (SCAT) system philosophy and benefits. IEEE Transactions on Vehicular...
  • Sutton, R. S., Barto, A. G., 2018. Reinforcement Learning: An Introduction, 2nd Edition. The MIT Press.
  • Tomar, I., Indu, S., Pandey, N., 2022. Traffic signal control methods: Current status, challenges, and emerging trends. In: Proceedings of...
  • TomTom International BV, 2025. Traffic index ranking. https://www.tomtom.com/traffic-index/ranking/
  • United Nations, 2018. World urbanization prospects: The 2018 revision. Tech. rep., United Nations, Department of Economic and Social Affairs.
  • Zaghal, R., Thabatah, K., Salah, S., 2018. Towards a smart intersection using traffic load balancing algorithm. Proceedings of Computing Conference...