[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-ai-model-lets-underwater-robots-predict-drift-not-just-images":10,"sections":41},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":36,"feedback":40,"feedback_at":22,"cost_usd":40,"total_tokens":40},8728,"ai-model-lets-underwater-robots-predict-drift-not-just-images","AI Model Lets Underwater Robots Predict Drift, Not Just Images","A new arXiv preprint describes AquaWAM, a compact AI model that predicts an underwater robot's drift and inertia, not just future camera frames.","Robots that swim have to think about physics long after they stop moving, and a new AI model is finally accounting for that.\n\nResearchers describe AquaWAM in an arXiv preprint (arXiv:2609.33299, posted September 30, 2026 as a replace-cross submission to cs.AI), calling it the first World Action Model built specifically for underwater robots. Unlike prior World Action Models, which predict how a scene will look after a robot's next move, AquaWAM skips video prediction and instead models the physical forces that keep acting on a submerged vehicle after a command ends, including thruster dead band, inertial glide, and ambient currents, sensed via a doppler velocity log, inertial measurement unit, pressure sensor, and joint encoders, with cameras reserved for reading goals and target poses rather than predicting imagery. On the USIM benchmark's 20 underwater tasks, the paper reports a 72.6% success rate and decisions 2.7x faster than a prior method called U0 on an Nvidia Jetson AGX Orin chip; with the velocity sensor disabled, AquaWAM still hit 61.6% success versus U0's 39.4%.\n\nThat distinction matters because most embodied-AI world models, tuned on wheeled or legged robots and drones, treat physics as something the next image will show you. Underwater vehicles do not get that luxury: currents and buoyancy keep nudging a vehicle around long after the thrusters stop, and murky water makes vision an unreliable narrator anyway. By modeling compact navigation state instead of predicting pixels, AquaWAM also sidesteps the computational overhead that makes video-based world models impractical on the modest processors that actually fit inside a submersible.\n\nEvery number here comes from one team's own paper on one simulated benchmark, not an independent trial in open water, so treat the 72.6% figure as a lab result rather than a guarantee for the next flooded pipe inspection or reef survey.","[\"robotics\",\"underwater robotics\",\"ai research\",\"world models\"]","2026-09-30T04:00:00.000Z","2026-09-30T22:15:01.915Z","2026-09-30T22:15:07.640Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Add explicit attribution — name the source (arXiv preprint 2609.33299, replace-cross posting) and identify the researchers\u002Finstitution behind AquaWAM — since every stat currently reads as an unsourced bare fact with no named paper, authors, or publication date.","resolved","ai",[32,33,34,35],"robotics","underwater robotics","ai research","world models",[37],{"name":38,"url":39},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.33299",0,{"sections":42},[43,47,51,55,60,65,69,74,79,83,88,93,98,103],{"name":44,"slug":30,"count":45,"latest_published_at":46},"AI",5629,"2026-10-01T04:00:00.000Z",{"name":48,"slug":49,"count":50,"latest_published_at":46},"Security","security",816,{"name":52,"slug":53,"count":54,"latest_published_at":46},"Policy","policy",430,{"name":56,"slug":57,"count":58,"latest_published_at":59},"Deals","deals",298,"2026-09-30T21:00:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Hardware","hardware",196,"2026-09-30T13:00:00.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":46},"Science","science",163,{"name":70,"slug":71,"count":72,"latest_published_at":73},"Consumer Tech","consumer-tech",149,"2026-09-30T22:57:11.000Z",{"name":75,"slug":76,"count":77,"latest_published_at":78},"Dev Tools","dev-tools",93,"2026-10-01T02:30:48.000Z",{"name":80,"slug":81,"count":77,"latest_published_at":82},"Software","software","2026-09-30T21:41:11.000Z",{"name":84,"slug":85,"count":86,"latest_published_at":87},"Startups","startups",84,"2026-09-30T20:39:09.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"Gaming","gaming",51,"2026-09-30T16:24:30.000Z",{"name":94,"slug":95,"count":96,"latest_published_at":97},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":99,"slug":100,"count":101,"latest_published_at":102},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":104,"slug":105,"count":106,"latest_published_at":107},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]