{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,22]],"date-time":"2026-09-22T07:39:22Z","timestamp":1790062762847,"version":"4.0.1"},"reference-count":30,"publisher":"MDPI AG","issue":"23","license":[{"start":{"date-parts":[[2025,12,2]],"date-time":"2025-12-02T00:00:00Z","timestamp":1764633600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"The Shanghai Maritime University Research Fund \u201cDevelopment and Application of New Technologies for Intelligent Shipping and Safety Management of Shipping Companies\u201d","award":["H20230487"],"award-info":[{"award-number":["H20230487"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>The autonomous coordination of multi-robot systems in complex, environments remains a fundamental challenge. Current Multi-Agent Reinforcement Learning (MARL) methods often struggle to reason effectively about the dynamic, causal relationships between agents and their surroundings. To address this, we introduce the Graph-Gated Transformer (GGT), a novel neural architecture designed to inject explicit relational priors directly into the self-attention mechanism for multi-robot coordination. The core mechanism of the GGT involves dynamically constructing a Tactical Relational Graph that encodes high-priority relationships like collision risk and cooperative intent. This graph is then used to generate an explicit attention mask, compelling the Transformer to focus its reasoning exclusively on entities rather than engaging in brute-force pattern matching across all perceived objects. Integrated into a Centralized Training with Decentralized Execution (CTDE) framework with QMIX, our approach demonstrates substantial improvements in high-fidelity simulations. In complex scenarios with dynamic obstacles and sensor noise, our GGT-based system achieves 95.3% coverage area efficiency with only 0.4 collisions per episode, a stark contrast to the 60.3% coverage and 20.7 collisions of standard QMIX. Ablation studies confirm that this structured, gated attention mechanism\u2014not merely the presence of attention\u2014is the key to unlocking robust collective autonomy. This work establishes that explicitly constraining the Transformer\u2019s attention space with dynamic, domain-aware relational graphs is a powerful and effective architectural solution for engineering safe and intelligent multi-robot systems.<\/jats:p>","DOI":"10.3390\/s25237335","type":"journal-article","created":{"date-parts":[[2025,12,5]],"date-time":"2025-12-05T18:42:02Z","timestamp":1764960122000},"page":"7335","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":2,"title":["Graph-Gated Relational Reasoning for Enhanced Coordination and Safety in Distributed Multi-Robot Systems: A Decentralized Reinforcement Learning Approach"],"prefix":"10.3390","volume":"25","author":[{"given":"Tianshun","family":"Chang","sequence":"first","affiliation":[{"name":"Merchant Shipping Academy, Shanghai Maritime University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yiping","family":"Ma","sequence":"additional","affiliation":[{"name":"Merchant Shipping Academy, Shanghai Maritime University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhiqian","family":"Li","sequence":"additional","affiliation":[{"name":"Merchant Shipping Academy, Shanghai Maritime University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shuai","family":"Huang","sequence":"additional","affiliation":[{"name":"College of Energy and Mechanical Engineering, Shanghai Electric Power University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zeqi","family":"Ma","sequence":"additional","affiliation":[{"name":"Merchant Shipping Academy, Shanghai Maritime University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yang","family":"Xiong","sequence":"additional","affiliation":[{"name":"Engineering College, Shanghai Ocean University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shijie","family":"Huang","sequence":"additional","affiliation":[{"name":"Engineering College, Shanghai Ocean University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jingbo","family":"Qin","sequence":"additional","affiliation":[{"name":"Engineering College, Shanghai Ocean University, Shanghai 201306, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2025,12,2]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Liu, Y., Sun, Z., Wan, J., Li, H., Yang, D., Li, Y., Fu, W., Yu, Z., and Sun, J. (2025). Hybrid Path Planning Method for USV Based on Improved A-Star and DWA. J. Mar. Sci. Eng., 13.","DOI":"10.3390\/jmse13050934"},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Shem-Tov, E., Sipper, M., and Elyasaf, A. (2025). BERT Mutation: Deep Transformer Model for Masked Uniform Mutation in Genetic Programming. Mathematics, 13.","DOI":"10.3390\/math13050779"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Li, Y., Li, L., Zhu, L., Zhang, Z., and Guo, Y. (2025). Leader-Following-Based Optimal Fault-Tolerant Consensus Control for Air\u2013Marine\u2013Submarine Heterogeneous Systems. J. Mar. Sci. Eng., 13.","DOI":"10.3390\/jmse13050878"},{"key":"ref_4","first-page":"1","article-title":"Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning","volume":"21","author":"Rashid","year":"2020","journal-title":"J. Mach. Learn. Res."},{"key":"ref_5","doi-asserted-by":"crossref","unstructured":"Manzini, T., and Murphy, R. (2024, January 13\u201317). Differentiable Boustrophedon Paths That Enable Optimization Via Gradient Descent. Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), Yokohama, Japan.","DOI":"10.1109\/ICRA57147.2024.10610136"},{"key":"ref_6","first-page":"840","article-title":"Multi-Task Offloading for Digital Twin-Enabled IoT Systems: A Deep Reinforcement Learning Approach","volume":"19","author":"Dai","year":"2023","journal-title":"IEEE Trans. Ind. Inform."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"1738","DOI":"10.1109\/JPROC.2019.2918951","article-title":"Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing","volume":"107","author":"Zhou","year":"2019","journal-title":"Proc. IEEE"},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Fossen, T.I. (2011). Handbook of Marine Craft Hydrodynamics and Motion Control, John Wiley & Sons.","DOI":"10.1002\/9781119994138"},{"key":"ref_9","doi-asserted-by":"crossref","unstructured":"Chen, S., Zhang, Z., Yang, Y., and Du, Y. (2024, January 20\u201327). STAS: Spatial-Temporal Return Decomposition for Multi-Agent Reinforcement Learning. Proceedings of the 38th Annual AAAI Conference on Artificial Intelligence, Vancouver, BC, Canada.","DOI":"10.1609\/aaai.v38i16.29681"},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"1582","DOI":"10.1109\/TVT.2023.3312574","article-title":"Stochastic Graph Neural Network-Based Value Decomposition for MARL in Internet of Vehicles","volume":"73","author":"Xiao","year":"2023","journal-title":"IEEE Trans. Veh. Technol."},{"key":"ref_11","doi-asserted-by":"crossref","unstructured":"Kramar, V., Dementiev, K., and Kabanov, A. (2025). Optimal State Estimation in Underwater Vehicle Discrete-Continuous Measurements via Augmented Hybrid Kalman Filter. J. Mar. Sci. Eng., 13.","DOI":"10.3390\/jmse13050933"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Bijelic, M., Gruber, T., Mannan, F., Kraus, F., Ritter, W., Dietmayer, K., and Heide, F. (2020, January 13\u201319). Seeing Through Fog Without Seeing Fog: Deep Multi-Modal Sensor Fusion in Unseen Adverse Weather. Proceedings of the 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, WA, USA.","DOI":"10.1109\/CVPR42600.2020.01170"},{"key":"ref_13","unstructured":"Sunehag, P., Lever, G., Gruslys, A., Czarnecki, W.M., Zambaldi, V., Jaderberg, M., Lanctot, M., Sonnerat, N., Leibo, J.Z., and Tuyls, K. (2018, January July). Value-Decomposition Networks for Cooperative Multi-Agent Learning. Proceedings of the 17th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2018), Stockholm, Sweden."},{"key":"ref_14","unstructured":"Son, K., Kim, D., Kang, W.J., Hostallero, D., and Yi, Y. (2019, January 10\u201315). QTRAN: Learning to Factorize with Transformation for Cooperative Multi-Agent Reinforcement Learning. Proceedings of the 36th International Conference on Machine Learning, Long Beach, CA, USA."},{"key":"ref_15","unstructured":"Son, K., Ahn, S., Reyes, R.D., Kim, D., and Kang, W.J. (2020). QTRAN++: Improved Value Transformation for Cooperative Multi-Agent Reinforcement Learning. arXiv."},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Cui, Z., Deng, K., Zhang, H., Zha, Z., and Jobaer, S. (2025). Deep Reinforcement Learning-Based Multi-Agent System with Advanced Actor\u2013Critic Framework for Complex Environment. Mathematics, 13.","DOI":"10.3390\/math13050754"},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Wu, N., Li, J., and Xiong, J. (2025). An Integrated Design of Course-Keeping Control and Extended State Observers for Nonlinear USVs with Disturbances. J. Mar. Sci. Eng., 13.","DOI":"10.3390\/jmse13050967"},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Hao, Y., Song, S., Huang, B., and Li, J. (2022, January 27\u201330). Distributed Edge-Event Triggered Formation Control for Multiple Unmanned Surface Vessels with Connectivity Preservation. Proceedings of the 2022 IEEE 17th International Conference on Control & Automation (ICCA), Hefei, China.","DOI":"10.1109\/ICCA54724.2022.9831864"},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"4559","DOI":"10.1109\/TNNLS.2022.3183865","article-title":"Residual Q-Networks for Value Function Factorizing in Multi-Agent Reinforcement Learning","volume":"35","author":"Pina","year":"2024","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"Memarian, F., Goo, W., Lioutikov, R., Niekum, S., and Topcu, U. (October, January 27). Self-Supervised Online Reward Shaping in Sparse-Reward Environments. Proceedings of the 2021 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS), Prague, Czech Republic.","DOI":"10.1109\/IROS51168.2021.9636020"},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Borgianni, L., Adami, D., Giordano, S., and Pagano, M. (2024). Enhancing Reliability in Rural Networks Using a Software-Defined Wide Area Network. Computers, 13.","DOI":"10.3390\/computers13050113"},{"key":"ref_22","first-page":"2792","article-title":"Constraint-Driven Multi-USV Coverage Path Generation for Aquatic Environmental Monitoring","volume":"31","author":"Toyomoto","year":"2023","journal-title":"IEEE Trans. Control Syst. Technol."},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Ni, J., Gu, Y., and Tang, G. (2024). Cooperative Coverage Path Planning for Multi-Mobile Robots Based on Improved K-Means Clustering and Deep Reinforcement Learning. Electronics, 13.","DOI":"10.3390\/electronics13050944"},{"key":"ref_24","unstructured":"De Vries, J.A., Moerland, T.M., and Plaat, A. (2022). On Credit Assignment in Hierarchical Reinforcement Learning. arXiv."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Zhang, J., Zhou, W., and Deng, X. (2025). Optimization of Adaptive Observation Strategies for Multi-AUVs in Complex Marine Environments Using Deep Reinforcement Learning. J. Mar. Sci. Eng., 13.","DOI":"10.3390\/jmse13050865"},{"key":"ref_26","unstructured":"Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, \u0141., and Polosukhin, I. (2017, January 4\u20139). Attention Is All You Need. Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, CA, USA."},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Ke, H., Wang, H., and Sun, H. (2022). Multi-Agent Deep Reinforcement Learning-Based Partial Task Offloading and Resource Allocation in Edge Computing Environment. Electronics, 11.","DOI":"10.3390\/electronics11152394"},{"key":"ref_28","unstructured":"Hu, S., Zhu, F., Chang, X., and Liang, X. (2021, January 3\u20137). UPDeT: Universal Multi-Agent Reinforcement Learning via Policy Decoupling with Transformers. Proceedings of the 9th International Conference on Learning Representations, Virtual Event."},{"key":"ref_29","doi-asserted-by":"crossref","unstructured":"Dai, Y., Zhang, Y., Zhou, X., Wang, Q., Song, X., and Wang, S. (2025). MTGNet: Multi-Agent End-to-End Motion Trajectory Prediction with Multi-Modal Panoramic Dynamic Graph. Appl. Sci., 15.","DOI":"10.3390\/app15105244"},{"key":"ref_30","doi-asserted-by":"crossref","unstructured":"Xu, Z., Shen, Y., Xie, Z., and Liu, Y. (2024). Research on Autonomous Underwater Vehicle Path Optimization Using a Field Theory-Guided A Algorithm. J. Mar. Sci. Eng., 12.","DOI":"10.3390\/jmse12101815"}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/25\/23\/7335\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,5]],"date-time":"2025-12-05T18:47:08Z","timestamp":1764960428000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/25\/23\/7335"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,2]]},"references-count":30,"journal-issue":{"issue":"23","published-online":{"date-parts":[[2025,12]]}},"alternative-id":["s25237335"],"URL":"https:\/\/doi.org\/10.3390\/s25237335","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,2]]}}}