{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,18]],"date-time":"2026-09-18T06:09:41Z","timestamp":1789711781766,"version":"4.0.1"},"reference-count":50,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,18]],"date-time":"2026-09-18T00:00:00Z","timestamp":1789689600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"<jats:p>Existing multi-agent reinforcement learning (MARL) approaches for traffic signal control often fail to capture two critical characteristics of real-world urban traffic: the dynamic spatio-temporal propagation of congestion across intersections and the frequent imperfection of sensor data. To address these fundamental gaps, this paper introduces hybrid coordinated spatio-temporal graph reinforcement learning (HC-STGRL), a novel framework that integrates three complementary mechanisms-gated recurrent unit (GRU), graph attention network (GAT), dueling double deep Q-Network (3DQN). Each intersection is equipped with a GRU to retain and update short-term memory of evolving traffic patterns, while a GAT enables selective and adaptive information sharing with neighboring intersections, allowing agents to weigh the relevance of spatio-temporal dependencies from different sources. The final phase decisions of the signal are derived using 3DQN. The proposed system is extensively evaluated using SUMO simulations on two synthetic grid networks and two real-world traffic networks. Experimental results demonstrate that HC-STGRL consistently reduces average trip times and delays compared to state-of-the-art MARL baselines, achieving up to a 11.3% reduction in travel time and a 14.9% reduction in waiting time relative to the strongest baseline, CoLight, on real-world road networks. Furthermore, the framework exhibits notable robustness, maintaining efficient operation even when 20% of sensors malfunction, a condition that significantly degrades the performance of existing methods: HC-STGRL incurs only a 25.4% performance degradation under 20% sensor failure, compared with 44.7% for CoLight. Importantly, HC-STGRL incorporates built-in mechanisms to prevent major arterial roads from starving adjacent intersections, thereby overcoming a well-documented limitation of conventional throughput-maximizing control strategies.<\/jats:p>","DOI":"10.3389\/frai.2026.1891950","type":"journal-article","created":{"date-parts":[[2026,9,18]],"date-time":"2026-09-18T05:37:54Z","timestamp":1789709874000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Beyond static snapshots: gated recurrent units and graph attention networks for smarter multi-agent traffic control"],"prefix":"10.3389","volume":"9","author":[{"family":"Mohita","sequence":"first","affiliation":[{"name":"School of Computer Science and Engineering (SCOPE), Vellore Institute of Technology (VIT)","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sidak Deep","family":"Singh","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering (SCOPE), Vellore Institute of Technology (VIT)","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"D.","family":"Subitha","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering (SCOPE), Vellore Institute of Technology (VIT)","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"J. C.","family":"Kavitha","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering (SCOPE), Vellore Institute of Technology (VIT)","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,18]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"278","DOI":"10.1061\/(ASCE)0733-947X(2003)129:3(278)","article-title":"Reinforcement learning for true adaptive traffic signal control","volume":"129","author":"Abdulhai","year":"2003","journal-title":"J. Transp. Eng"},{"key":"B2","doi-asserted-by":"publisher","first-page":"128","DOI":"10.1049\/iet-its.2009.0070","article-title":"Reinforcement learning-based multi-agent system for network traffic signal control","volume":"4","author":"Arel","year":"2010","journal-title":"IET Intell. Transp. Syst"},{"key":"B3","article-title":"\u201cReinforcement learning benchmarks for traffic signal control,\u201d","author":"Ault","year":"2021"},{"key":"B4","doi-asserted-by":"publisher","first-page":"156","DOI":"10.1109\/TSMCC.2007.913919","article-title":"\u201cA comprehensive survey of multiagent reinforcement learning.\u201d","author":"Busoniu","year":"2008","journal-title":"IEEE transactions on systems, man, and cybernetics, part C"},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2605.10170","article-title":"Balancing efficiency and fairness in traffic light control through deep reinforcement learning","author":"Cederle","year":"2026","journal-title":"arXiv"},{"key":"B6","first-page":"2746","article-title":"\u201cPressLight: learning max pressure control to coordinate traffic signals in arterial network,\u201d","author":"Chen","year":"2019"},{"key":"B7","doi-asserted-by":"publisher","first-page":"3414","DOI":"10.1609\/aaai.v34i04.5744","article-title":"\u201cToward a thousand lights: decentralized deep reinforcement learning for large-scale traffic signal control,\u201d","volume":"34","author":"Chen","year":"2020","journal-title":"Proceedings of the AAAI conference on artificial intelligence"},{"key":"B8","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2206.11996","article-title":"The real deal: a review of challenges and opportunities in moving reinforcement learning-based traffic signal control systems towards reality","author":"Chen","year":"2022","journal-title":"arXiv"},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.1007\/s10723-024-09762-6","article-title":"Exploring the synergy of blockchain, IoT, and edge computing in smart traffic management across urban landscapes","author":"Chen","year":"2024","journal-title":"J. Grid Comput"},{"key":"B10","doi-asserted-by":"publisher","first-page":"1724","DOI":"10.3115\/v1\/D14-1179","article-title":"\u201cLearning phrase representations using rnn encoder-decoder for statistical machine translation,\u201d","author":"Cho","year":"2014"},{"key":"B11","doi-asserted-by":"publisher","first-page":"1086","DOI":"10.1109\/TITS.2019.2901791","article-title":"\u201cMulti-agent deep reinforcement learning for large-scale traffic signal control,\u201d","author":"Chu","year":"2020","journal-title":"IEEE transactions on intelligent transportation systems"},{"key":"B12","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.3555","article-title":"Empirical evaluation of gated recurrent neural networks on sequence modeling","author":"Chung","year":"2014","journal-title":"arXiv"},{"key":"B13","doi-asserted-by":"publisher","first-page":"1140","DOI":"10.1109\/TITS.2013.2255286","article-title":"Multi-agent reinforcement learning for integrated network of adaptive traffic signal controllers (MARLIN-ATSC)","volume":"14","author":"El-Tantawy","year":"2013","journal-title":"IEEE Trans. Intell. Transp. Syst."},{"key":"B14","doi-asserted-by":"publisher","first-page":"105","DOI":"10.1007\/978-3-319-15024-6_7","article-title":"\u201cSumo's lane-changing model,\u201d","author":"Erdmann","year":"2015","journal-title":"Modeling Mobility with Open Data"},{"key":"B15","volume-title":"Manual on Uniform Traffic Control Devices for Streets and Highways","year":"2009"},{"key":"B16","doi-asserted-by":"publisher","first-page":"107","DOI":"10.1142\/S0218488598000094","article-title":"Vanishing gradient problem","volume":"6","author":"Hochreiter","year":"1998","journal-title":"Int. J. Uncertain. Fuzziness"},{"key":"B17","doi-asserted-by":"publisher","first-page":"3911","DOI":"10.1111\/mice.70037","article-title":"Explainable reinforcement learning for improved traffic signal control","volume":"40","author":"Hu","year":"2025","journal-title":"Comput. Aided Civ. Infrastruct. Eng"},{"key":"B18","author":"Hunt","year":"1981","journal-title":"SCOOT"},{"key":"B19","doi-asserted-by":"crossref","DOI":"10.24963\/ijcai.2024\/11","article-title":"\u201cX-light: cross-city traffic signal control using transformer on transformer as meta multi-agent reinforcement learner,\u201d","author":"Jiang","year":"2024"},{"key":"B20","article-title":"\u201cAdam: a method for stochastic optimization,\u201d","author":"Kingma","year":"2015"},{"key":"B21","volume-title":"Microscopic modeling of traffic flow: investigation of collision free vehicle dynamics","author":"Krau\u00df","year":"1998"},{"key":"B22","doi-asserted-by":"publisher","first-page":"652","DOI":"10.1109\/CASE48305.2020.9216899","article-title":"\u201cFairness control of traffic light via deep reinforcement learning,\u201d","author":"Li","year":"2020"},{"key":"B23","author":"Lieberman","year":"2000","journal-title":"Traffic Signal Control Systems"},{"key":"B24","doi-asserted-by":"publisher","first-page":"317","DOI":"10.1098\/rspa.1955.0089","article-title":"\u201cOn kinematic waves. II. A theory of traffic flow on long crowded roads,\u201d","volume":"229","author":"Lighthill","year":"1955"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.1109\/ITSC.2018.8569938","article-title":"\u201cMicroscopic traffic simulation using SUMO,\u201d","author":"Lopez","year":"2018"},{"key":"B26","article-title":"Scats, Sydney coordinated adaptive traffic system: A traffic responsive method of controlling urban traffic","author":"Lowrie","year":"1990","journal-title":"Road Transp. Res"},{"key":"B27","doi-asserted-by":"publisher","first-page":"529","DOI":"10.1038\/nature14236","article-title":"Human-level control through deep reinforcement learning","volume":"518","author":"Mnih","year":"2015","journal-title":"Nature"},{"key":"B28","doi-asserted-by":"publisher","first-page":"36737","DOI":"10.1109\/ACCESS.2026.3671697","article-title":"\u201cIntegrated diagnosis and prognosis of dynamic systems using deep learning: case study on a welding robot,\u201d","author":"Ngnassi Djami","year":"2026"},{"key":"B29","doi-asserted-by":"publisher","first-page":"111485","DOI":"10.1016\/j.cie.2025.111485","article-title":"Meta-modeling of maintenance assets using an evolutionary algorithm approach: the case of a corn sheller","volume":"209","author":"Ngnassi Djami","year":"2025","journal-title":"Comput. Ind. Eng"},{"key":"B30","doi-asserted-by":"publisher","first-page":"2043","DOI":"10.1109\/JPROC.2003.819610","article-title":"\u201cReview of road traffic control strategies,\u201d","author":"Papageorgiou","year":"2003"},{"key":"B31","author":"Schrank","year":"2021","journal-title":"2021 Urban Mobility Report"},{"key":"B32","doi-asserted-by":"publisher","first-page":"263","DOI":"10.1007\/s42421-020-00029-6","article-title":"Robust deep reinforcement learning for traffic signal control","volume":"2","author":"Tan","year":"2020","journal-title":"J. Big Data Anal. Transp"},{"key":"B33","doi-asserted-by":"publisher","first-page":"1061","DOI":"10.1109\/TMC.2013.27","article-title":"\u201cGeneration and analysis of a large-scale urban vehicular mobility dataset,\u201d","author":"Uppoor","year":"2014","journal-title":"IEEE transactions on mobile computing"},{"key":"B34","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v30i1.10295","article-title":"\u201cDeep reinforcement learning with double q-learning,\u201d","author":"van Hasselt","year":"2016","journal-title":"Proceedings of the AAAI conference on artificial intelligence"},{"key":"B35","doi-asserted-by":"publisher","first-page":"435","DOI":"10.1007\/978-1-4614-6243-9_2","article-title":"The max-pressure controller for arbitrary networks of signalized intersections","volume":"7","author":"Varaiya","year":"2013","journal-title":"Ad. Automot. Control"},{"key":"B36","article-title":"\u201cGraph attention networks,\u201d","author":"Veli\u010dkovi\u0107","year":"2018"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2404.11014","article-title":"Towards multi-agent reinforcement learning based traffic signal control through spatio-temporal hypergraphs","author":"Wang","year":"2024","journal-title":"arXiv"},{"key":"B38","first-page":"1995","article-title":"\u201cDueling network architectures for deep reinforcement learning,\u201d","author":"Wang","year":"2016","journal-title":"International conference on machine learning"},{"key":"B39","author":"Webster","year":"1958","journal-title":"Traffic Signal Settings"},{"key":"B40","doi-asserted-by":"publisher","DOI":"10.1145\/3357384.3357902","article-title":"\u201cCoLight: learning network-level cooperation for traffic signal control,\u201d","author":"Wei","year":"2019"},{"key":"B41","doi-asserted-by":"publisher","first-page":"2496","DOI":"10.1145\/3219819.3220096","article-title":"\u201cIntelliLight: a reinforcement learning approach for intelligent traffic light control,\u201d","author":"Wei","year":"2018"},{"key":"B42","doi-asserted-by":"publisher","first-page":"2639","DOI":"10.1145\/3580305.3599530","article-title":"\u201cTransformerlight: a novel sequence modeling based traffic signaling mechanism via gated transformer,\u201d","author":"Wu","year":"2023"},{"key":"B43","doi-asserted-by":"publisher","first-page":"liaf016","DOI":"10.1093\/iti\/liaf009","article-title":"Advances in reinforcement learning for traffic signal control: a review of recent progress","volume":"4","author":"Xiao","year":"2025","journal-title":"Intell. Transp. Infrastruct"},{"key":"B44","doi-asserted-by":"publisher","first-page":"128","DOI":"10.1007\/s10462-026-11530-9","article-title":"Intelligent traffic signal control based on reinforcement learning: a survey","volume":"59","author":"Xiao","year":"2026","journal-title":"Artif. Intell. Rev"},{"key":"B45","doi-asserted-by":"publisher","first-page":"2238","DOI":"10.3390\/electronics15112238","article-title":"Adaptive spatio-temporal self-supervised traffic flow prediction method based on contrastive learning","volume":"15","author":"Xing","year":"2026","journal-title":"Electronics"},{"key":"B46","doi-asserted-by":"publisher","first-page":"880","DOI":"10.1177\/03611981241297979","article-title":"Multi-agent deep reinforcement learning with graph attention network for traffic signal control in multiple-intersection urban areas","volume":"2679","author":"Yang","year":"2025","journal-title":"Transp. Res. Rec. J. Transp. Res. Board"},{"key":"B47","doi-asserted-by":"publisher","DOI":"10.24963\/ijcai.2018\/505","article-title":"\u201cSpatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting,\u201d","author":"Yu","year":"2018","journal-title":"Proceedings of the 27th international joint conference on artificial intelligence"},{"key":"B48","doi-asserted-by":"publisher","first-page":"1153","DOI":"10.1609\/aaai.v34i01.5467","article-title":"\u201cMetaLight: value-based meta-reinforcement learning for traffic signal control,\u201d","volume":"34","author":"Zang","year":"2020","journal-title":"Proceedings of the AAAI conference on artificial intelligence"},{"key":"B49","doi-asserted-by":"publisher","first-page":"686","DOI":"10.3390\/s26020686","article-title":"Impact analysis of the market penetration rate of connected vehicles and the failure rate of roadside equipment on data accuracy","volume":"26","author":"Zhan","year":"2026","journal-title":"Sensors"},{"key":"B50","doi-asserted-by":"publisher","first-page":"3620","DOI":"10.1145\/3308558.3314139","article-title":"\u201cCityflow: a multi-agent reinforcement learning environment for large scale city traffic scenario,\u201d","author":"Zhang","year":"2019"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1891950\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,18]],"date-time":"2026-09-18T05:37:58Z","timestamp":1789709878000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1891950\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,18]]},"references-count":50,"alternative-id":["10.3389\/frai.2026.1891950"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1891950","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,18]]},"article-number":"1891950"}}