{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"Authorea Inc."}],"indexed":{"date-parts":[[2025,10,3]],"date-time":"2025-10-03T20:12:54Z","timestamp":1759522374773,"version":"build-2065373602"},"posted":{"date-parts":[[2025,10,3]]},"group-title":"Preprints","reference-count":0,"publisher":"Wiley","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2025,10,3]]},"abstract":"<jats:p id=\"p1\">Autonomous landing for Unmanned Aerial Vehicles (UAVs) requires both\nprecision and resilience against environmental uncertainties,\ncapabilities that current approaches struggle to deliver. This paper\npresents a novel learning-based solution that combines an advanced\nmultimodal transformer-based detector with a reinforcement learning\nformulation to achieve reliable autonomous landing behavior across\nvarying scenario uncertainties. Beyond the integration of multimodality\nfor robust target detection, this research incorporates a comprehensive\nanalysis of the impact of state representation on decision-making\nperformance. The proposed methodology is validated through extensive\nsimulation studies and real-world field experiments conducted on\nphysical UAV platforms under natural wind disturbances, demonstrating\nreliable transfer from simulated training environments to controlled\noutdoor conditions. Field experiments across varying initial conditions\nand wind stress confirm the system\u2019s robustness, achieving landing\nprecision of 0.10 \u00b1 0.08 meters in outdoor trials, demonstrating\ncentimeter-level accuracy that surpasses the meter-level precision of\nglobal positioning systems.<\/jats:p>","DOI":"10.22541\/au.175952136.69790680\/v1","type":"posted-content","created":{"date-parts":[[2025,10,3]],"date-time":"2025-10-03T19:56:16Z","timestamp":1759521376000},"source":"Crossref","is-referenced-by-count":0,"title":["A Multimodal Agentic AI for the Autonomous Precise Landing of UAVs"],"prefix":"10.22541","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4366-6227","authenticated-orcid":true,"given":"Francisco Soares Pinto","family":"Neves","sequence":"first","affiliation":[{"name":"Instituto de Engenharia de Sistemas e Computadores Tecnologia e Ciencia"}]},{"given":"Lu\u00eds Miguel","family":"Branco","sequence":"additional","affiliation":[{"name":"Instituto de Engenharia de Sistemas e Computadores Tecnologia e Ciencia"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8010-7507","authenticated-orcid":true,"given":"Rafael","family":"Claro","sequence":"additional","affiliation":[{"name":"Instituto de Engenharia de Sistemas e Computadores Tecnologia e Ciencia"}]},{"given":"Andry M.","family":"Pinto","sequence":"additional","affiliation":[{"name":"Universidade do Porto Faculdade de Engenharia"}]}],"member":"311","container-title":[],"original-title":[],"deposited":{"date-parts":[[2025,10,3]],"date-time":"2025-10-03T19:56:16Z","timestamp":1759521376000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.authorea.com\/users\/797642\/articles\/1341891-a-multimodal-agentic-ai-for-the-autonomous-precise-landing-of-uavs?commit=6998ffa51c420a87ebaa23be1ffde18d52e0ab7c"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,3]]},"references-count":0,"URL":"https:\/\/doi.org\/10.22541\/au.175952136.69790680\/v1","relation":{},"subject":[],"published":{"date-parts":[[2025,10,3]]},"subtype":"preprint"}}