{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,26]],"date-time":"2026-09-26T14:00:25Z","timestamp":1790431225148,"version":"4.1.0"},"reference-count":48,"publisher":"Wiley","issue":"1","license":[{"start":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T00:00:00Z","timestamp":1789430400000},"content-version":"vor","delay-in-days":14,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"tdm","delay-in-days":0,"URL":"http:\/\/doi.wiley.com\/10.1002\/tdm_license_1.1"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["71971220"],"award-info":[{"award-number":["71971220"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["71901093"],"award-info":[{"award-number":["71901093"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["nyaspubs.onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Ann NY Acad Sci"],"published-print":{"date-parts":[[2026,9]]},"abstract":"<jats:title>ABSTRACT<\/jats:title>\n                  <jats:p>Modern autonomous driving systems are constrained by the tension between high\u2010fidelity perception and stringent onboard energy conservation. We propose adaptive routing of multimodal onboard reasoning (ARMOR), an end\u2010to\u2010end heterogeneous mixture\u2010of\u2010experts framework that dynamically routes inputs between computationally efficient pure\u2010vision experts and robust radar\u2010vision fusion experts. A lightweight context\u2010aware gating network, the multimodal adaptive routing sentinel (MARS), integrates temporal context and environmental priors to select appropriate experts. ARMOR was optimized using a unified cost\u2010conditioned multiobjective loss, enabling a single model to learn routing policies across an empirical cost\u2013performance trade\u2010off frontier and adapt to different operational preferences. To improve reliability in high\u2010uncertainty scenarios, an evidential\u2010learning\u2010based sentry mechanism was used to trigger fallback to a high\u2010fidelity fusion expert. Experiments on the simulation\u2010generated CARLA\u2010SUMO multimodal object detection dataset (CSODD) showed that ARMOR\u2010safety retains 98.1% of the mean average precision of the strongest fusion baseline while reducing average floating\u2010point operations per second (FLOPs), latency, and the profiled GPU\u2010side inference\u2010energy proxy by 66.3%, 64.4%, and 65.8%, respectively. In adverse weather, it improved mean average precision by 58.3% over high\u2010performance pure\u2010vision models and reduced the critical\u2010object miss rate by more than 50%. These results provide controlled framework\u2010level evidence, but do not establish cross\u2010dataset generalization or real\u2010road deployment performance.<\/jats:p>","DOI":"10.1111\/nyas.70397","type":"journal-article","created":{"date-parts":[[2026,9,15]],"date-time":"2026-09-15T15:56:56Z","timestamp":1789487816000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["ARMOR: Pareto\u2010Aware Dynamic Routing for Energy\u2010Efficient Heterogeneous Multimodal Perception in Autonomous Driving"],"prefix":"10.1111","volume":"1563","author":[{"given":"Junli","family":"Liu","sequence":"first","affiliation":[{"name":"School of Traffic and Transportation Engineering Central South University  Changsha China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yinggui","family":"Zhang","sequence":"additional","affiliation":[{"name":"School of Traffic and Transportation Engineering Central South University  Changsha China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yanzhan","family":"Chen","sequence":"additional","affiliation":[{"name":"School of Traffic and Transportation Engineering Central South University  Changsha China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yue","family":"Wang","sequence":"additional","affiliation":[{"name":"School of Management Science and Engineering Stanford University  Stanford California USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"311","published-online":{"date-parts":[[2026,9,15]]},"reference":[{"key":"e_1_2_11_2_1","doi-asserted-by":"crossref","unstructured":"Y.Xia Z.Liang L.Deng Y.Zhao H.Su andK.Zheng \u201cEnergy\u2010Efficient Autonomous Driving with Adaptive Perception and Robust Decision \u201d arXiv preprint arXiv:2510.25205 2025.","DOI":"10.1109\/ICDE65706.2026.00057"},{"key":"e_1_2_11_3_1","doi-asserted-by":"crossref","unstructured":"Y.Chae H.Park H.Kim andK. J.Yoon \u201cDoppler\u2010Aware LiDAR\u2010RADAR Fusion for Weather\u2010Robust 3D Detection \u201d inProceedings of the IEEE\/CVF International Conference on Computer Vision(2025) 27197\u201327208.","DOI":"10.1109\/ICCV51701.2025.02525"},{"key":"e_1_2_11_4_1","doi-asserted-by":"publisher","DOI":"10.1109\/TIV.2024.3507538"},{"key":"e_1_2_11_5_1","doi-asserted-by":"crossref","unstructured":"A. V.Malawade T.Mortlock andM. A. A.Faruque \u201cEcoFusion: Energy\u2010Aware Adaptive Sensor Fusion for Efficient Autonomous Vehicle Perception \u201d inProceedings of the 59th ACM\/IEEE Design Automation Conference(2022) 481\u2013486.","DOI":"10.1145\/3489517.3530489"},{"key":"e_1_2_11_6_1","doi-asserted-by":"crossref","unstructured":"T.Chen X.Chen X.Du et\u00a0al. \u201cAdamv\u2010MoE: Adaptive Multi\u2010Task Vision Mixture\u2010of\u2010Experts \u201d inProceedings of the IEEE\/CVF International Conference on Computer Vision(2023) 17346\u201317357.","DOI":"10.1109\/ICCV51070.2023.01591"},{"key":"e_1_2_11_7_1","unstructured":"L.Kong J.Lin S.Li K.Luo Z.Li andK.Yang \u201cCoBEVMoE: Heterogeneity\u2010Aware Feature Fusion with Dynamic Mixture\u2010of\u2010Experts for Collaborative Perception.\u201d arXiv preprint arXiv:2509.17107 2025."},{"key":"e_1_2_11_8_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2025.114514"},{"key":"e_1_2_11_9_1","doi-asserted-by":"crossref","unstructured":"H.Hu L.Huang X.Wang et\u00a0al. \u201cNav\u2010EE: Navigation\u2010Guided Early Exiting for Efficient Vision\u2010Language Models in Autonomous Driving \u201d arXiv preprint arXiv:2510.01795 2025.","DOI":"10.1109\/RTCSA71153.2026.00019"},{"key":"e_1_2_11_10_1","unstructured":"K.Huang B.Shi X.Li X.Li S.Huang andY.Li \u201cMulti\u2010Modal Sensor Fusion for Auto Driving Perception: A Survey \u201d arXiv preprint arXiv:2202.02703 2022."},{"key":"e_1_2_11_11_1","doi-asserted-by":"publisher","DOI":"10.3390\/s25196033"},{"key":"e_1_2_11_12_1","doi-asserted-by":"crossref","unstructured":"S.Pang D.Morris andH.Radha \u201cTransCAR: Transformer\u2010Based Camera\u2010and\u2010Radar Fusion for 3D Object Detection \u201d in2023 IEEE\/RSJ International Conference on Intelligent Robots and Systems (IROS)(2023) 10902\u201310909.","DOI":"10.1109\/IROS55552.2023.10341793"},{"key":"e_1_2_11_13_1","doi-asserted-by":"crossref","unstructured":"J.Kim M.Seong G.Bang D.Kum andJ. W.Choi \u201cRCM\u2010Fusion: Radar\u2010Camera Multi\u2010Level Fusion for 3D Object Detection \u201d in2024 IEEE International Conference on Robotics and Automation (ICRA)(2024) 18236\u201318242.","DOI":"10.1109\/ICRA57147.2024.10611449"},{"key":"e_1_2_11_14_1","doi-asserted-by":"crossref","unstructured":"Z.Liu H.Tang A.Amini X.Yang H.Mao D.Rus andS.Han \u201cBEVFusion: Multi\u2010Task Multi\u2010Sensor Fusion with Unified Bird's\u2010Eye View Representation \u201d arXiv preprint arXiv:2205.13542 2022.","DOI":"10.1109\/ICRA48891.2023.10160968"},{"key":"e_1_2_11_15_1","article-title":"BEVFormer: Learning Bird's\u2010Eye\u2010View Representation From Lidar\u2010Camera via Spatiotemporal Transformers","author":"Li Z.","year":"2024","journal-title":"IEEE Transactions on Pattern Analysis and Machine Intelligence"},{"key":"e_1_2_11_16_1","doi-asserted-by":"crossref","unstructured":"Z.Lin Z.Liu Z.Xia et\u00a0al. \u201cRCBEVDet: Radar\u2010Camera Fusion in Bird's Eye View for 3D Object Detection \u201d inProceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition(2024) 14928\u201314937.","DOI":"10.1109\/CVPR52733.2024.01414"},{"key":"e_1_2_11_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2021.3117837"},{"key":"e_1_2_11_18_1","doi-asserted-by":"crossref","unstructured":"M.Sohan T.Sai Ram andC. V.Rami Reddy \u201cA Review on Yolov8 and its Advancements \u201d inInternational Conference on Data Intelligence and Cognitive Informatics(2024) 529\u2013545.","DOI":"10.1007\/978-981-99-7962-2_39"},{"key":"e_1_2_11_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/LRA.2025.3536218"},{"key":"e_1_2_11_20_1","doi-asserted-by":"crossref","unstructured":"A. V.Malawade T.Mortlock andM. A.Al Faruque \u201cHydraFusion: Context\u2010Aware Selective Sensor Fusion for Robust and Efficient Autonomous Vehicle Perception \u201d in2022 ACM\/IEEE 13th International Conference on Cyber\u2010Physical Systems (ICCPS)(2022) 68\u201379.","DOI":"10.1109\/ICCPS54341.2022.00013"},{"key":"e_1_2_11_21_1","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2024.3435937"},{"key":"e_1_2_11_22_1","unstructured":"S.Rajbhandari C.Li Z.Yao et\u00a0al. \u201cDeepspeed\u2010MoE: Advancing Mixture\u2010of\u2010Experts Inference and Training to Power Next\u2010Generation AI Scale \u201d inInternational Conference on Machine Learning(2022) 18332\u201318346."},{"key":"e_1_2_11_23_1","unstructured":"B.Lin Z.Tang Y.Ye et\u00a0al. \u201cMoe\u2010llava: Mixture of Experts for Large Vision\u2010Language Models \u201d arXiv preprint arXiv:2401.15947 2024."},{"key":"e_1_2_11_24_1","unstructured":"A.Liu B.Feng B.Xue et\u00a0al. \u201cDeepseek\u2010v3 Technical Report \u201d arXiv preprint arXiv:2412.19437 2024."},{"key":"e_1_2_11_25_1","doi-asserted-by":"crossref","unstructured":"Q.Wang X.Han J.Wang et\u00a0al. \u201cMultiPL\u2010MoE: Multi\u2010Programming\u2010Lingual Extension of Large Language Models through Hybrid Mixture\u2010of\u2010Experts \u201din EMNLP (Findings) pp. 12817\u201012828. 2025.","DOI":"10.18653\/v1\/2025.findings-emnlp.686"},{"key":"e_1_2_11_26_1","unstructured":"Y.Wang M.Huo Z.Liang et\u00a0al. \u201cVER: Vision Expert Transformer for Robot Learning via Foundation Distillation and Dynamic Routing \u201d arXiv preprint arXiv:2510.05213 2025."},{"key":"e_1_2_11_27_1","unstructured":"Z.Li Y.Liu C.Hui J.Lee S.Lee andW.Lin \u201cShape Distribution Matters: Shape\u2010Specific Mixture\u2010of\u2010Experts for Amodal Segmentation under Diverse Occlusions \u201d arXiv preprint arXiv:2508.01664 2025."},{"key":"e_1_2_11_28_1","doi-asserted-by":"crossref","unstructured":"M.Liao H.Dong X.Wang K.Ubul Y.Shao andZ.Yan \u201cGM\u2010MoE: Low\u2010Light Enhancement With Gated\u2010Mechanism Mixture\u2010of\u2010Experts \u201d inProceedings of the IEEE\/CVF International Conference on Computer Vision(2025) 8766\u20138776.","DOI":"10.1109\/ICCV51701.2025.00820"},{"key":"e_1_2_11_29_1","unstructured":"Z.Yang Y.Chai X.Jia et\u00a0al. \u201cDriveMoE: Mixture\u2010of\u2010Experts for Vision\u2010Language\u2010Action Model in End\u2010to\u2010End Autonomous Driving \u201d arXiv preprint arXiv:2505.16278 2025."},{"key":"e_1_2_11_30_1","doi-asserted-by":"crossref","unstructured":"A.Howard M.Sandler G.Chu et\u00a0al. \u201cSearching for MobileNetV3 \u201d inProceedings of the IEEE\/CVF International Conference on Computer Vision(2019) 1314\u20131324.","DOI":"10.1109\/ICCV.2019.00140"},{"key":"e_1_2_11_31_1","unstructured":"C. R.Qi H.Su K.Mo andL. J.Guibas \u201cPointNet: Deep Learning on Point Sets for 3D Classification and Segmentation \u201d inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition(2017) 652\u2013660."},{"key":"e_1_2_11_32_1","doi-asserted-by":"publisher","DOI":"10.3390\/rs13030506"},{"key":"e_1_2_11_33_1","unstructured":"F. M.Shiri T.Perumal N.Mustapha andR.Mohamed \u201cA Comprehensive Overview and Comparative Analysis on Deep Learning Models: CNN RNN LSTM GRU \u201d arXiv preprint arXiv:2305.17473 2023."},{"key":"e_1_2_11_34_1","unstructured":"E.Jang S.Gu andB.Poole \u201cCategorical Reparameterization With Gumbel\u2010Softmax \u201d arXiv preprint arXiv:1611.01144 2016."},{"key":"e_1_2_11_35_1","first-page":"12454","article-title":"Argmax Flows and Multinomial Diffusion: Learning Categorical Distributions","volume":"34","author":"Hoogeboom E.","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_2_11_36_1","doi-asserted-by":"publisher","DOI":"10.1111\/nyas.12665"},{"key":"e_1_2_11_37_1","unstructured":"L. L.Folgoc V.Baltatzis S.Desai et\u00a0al. \u201cIs MC Dropout Bayesian? \u201d arXiv preprint arXiv:2110.04286 2021."},{"key":"e_1_2_11_38_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-42553-1_3"},{"key":"e_1_2_11_39_1","first-page":"3183","article-title":"Evidential Deep Learning to Quantify Classification Uncertainty","volume":"31","author":"Sensoy M.","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_2_11_40_1","first-page":"330","article-title":"Dempster\u2010Shafer Theory","volume":"1","author":"Shafer G.","year":"1992","journal-title":"Encyclopedia of Artificial Intelligence"},{"key":"e_1_2_11_41_1","unstructured":"L.Wang H.Gao C.Zhao X.Sun andD.Dai \u201cAuxiliary\u2010Loss\u2010Free Load Balancing Strategy for Mixture\u2010of\u2010Experts \u201d arXiv preprint arXiv:2408.15664 2024."},{"key":"e_1_2_11_42_1","doi-asserted-by":"publisher","DOI":"10.52202\/079017-3429"},{"key":"e_1_2_11_43_1","doi-asserted-by":"crossref","unstructured":"Y.Xia Z.Liang L.Deng Y.Zhao H.Su andK.Zheng \u201cEnergy\u2010Efficient Autonomous Driving With Adaptive Perception and Robust Decision \u201d arXiv preprint arXiv:2510.25205 2025.","DOI":"10.1109\/ICDE65706.2026.00057"},{"key":"e_1_2_11_44_1","first-page":"1","article-title":"Switch Transformers: Scaling to Trillion Parameter Models With Simple and Efficient Sparsity","volume":"23","author":"Fedus W.","year":"2022","journal-title":"Journal of Machine Learning Research"},{"key":"e_1_2_11_45_1","unstructured":"H.Cai C.Gan T.Wang Z.Zhang andS.Han Once\u2010for\u2010All: Train One Network and Specialize it for Efficient Deployment. arXiv preprint arXiv:1908.09791. (2019)."},{"key":"e_1_2_11_46_1","unstructured":"N.Carion L.Gustafson Y. T.Hu et al. \u201cSam 3: Segment Anything With Concepts \u201d arXiv preprint arXiv:2511.16719 2025."},{"key":"e_1_2_11_47_1","doi-asserted-by":"publisher","DOI":"10.1038\/s43586-021-00018-1"},{"key":"e_1_2_11_48_1","unstructured":"C.Lyu W.Zhang H.Huang et al. Rtmdet: An Empirical Study of Designing Real\u2010Time Object Detectors. arXiv preprint arXiv:2212.07784. (2022)."},{"key":"e_1_2_11_49_1","doi-asserted-by":"publisher","DOI":"10.3390\/s25133865"}],"container-title":["Annals of the New York Academy of Sciences"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/nyaspubs.onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/nyas.70397","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/nyaspubs.onlinelibrary.wiley.com\/doi\/full-xml\/10.1111\/nyas.70397","content-type":"application\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/nyaspubs.onlinelibrary.wiley.com\/doi\/pdf\/10.1111\/nyas.70397","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,26]],"date-time":"2026-09-26T13:12:27Z","timestamp":1790428347000},"score":1,"resource":{"primary":{"URL":"https:\/\/nyaspubs.onlinelibrary.wiley.com\/doi\/10.1111\/nyas.70397"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9]]},"references-count":48,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2026,9]]}},"alternative-id":["10.1111\/nyas.70397"],"URL":"https:\/\/doi.org\/10.1111\/nyas.70397","archive":["Portico"],"relation":{},"ISSN":["0077-8923","1749-6632"],"issn-type":[{"value":"0077-8923","type":"print"},{"value":"1749-6632","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9]]},"assertion":[{"value":"2026-01-25","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-08-05","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-09-15","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70397"}}