[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-robot-ai-models-get-45x-faster-by-rethinking-how-they-plan":10,"sections":40},{"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":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},7750,"robot-ai-models-get-45x-faster-by-rethinking-how-they-plan","Robot AI Models Get 4.5x Faster by Rethinking How They Plan","A new technique called Rolling-WAM speeds up robot planning by denoising action chunks gradually instead of all at once, cutting replanning lag 4.5x.","Robots that plan their next move by imagining video of the future just got a lot quicker at it.\n\nA paper posted to arXiv describes Rolling-WAM, a new way to run World Action Models, the systems that generate a robot's next actions alongside a predicted video of what those actions will look like. Normally, that video-and-action prediction has to be fully worked out from scratch before every single move, which creates lag. Rolling-WAM instead keeps a rolling window of upcoming action chunks at different stages of completion: the next move gets finished first, while moves further out stay partially sketched and get refined as the robot gets closer to needing them. Tested on the LIBERO and RoboTwin simulation benchmarks and on a real Unitree G1 humanoid, the approach matched standard methods on manipulation tasks while replanning 4.5 times faster.\n\nThe real story here is not the manipulation scores, it's the latency fix. Robots that predict video frame-by-frame before acting have always been stuck choosing between accuracy and speed, since full-horizon denoising is expensive to redo constantly. Treating the prediction window like a pipeline instead of a one-shot computation is a fairly standard trick in other areas of machine learning, and its arrival here suggests robotics is starting to borrow more systems-engineering tricks rather than just bigger models.\n\nA 4.5x speedup in a lab with a Unitree G1 is a promising number, not a guarantee it holds up in a cluttered warehouse or someone's kitchen.","[\"robotics\",\"ai\",\"world-models\",\"arxiv\"]","2026-09-25T04:00:00.000Z","2026-09-25T19:44:35.569Z","2026-09-25T19:44:41.236Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Remove the stray, empty bullet-point marker ('- ') left in the middle of the body between the two main paragraphs before publishing.","resolved","ai",[32,30,33,34],"robotics","world-models","arxiv",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.30247",0,{"sections":41},[42,46,51,56,61,66,71,76,81,86,91,96,101,106],{"name":43,"slug":30,"count":44,"latest_published_at":45},"AI",4482,"2026-09-25T15:40:03.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Security","security",734,"2026-09-25T15:52:13.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Policy","policy",388,"2026-09-25T15:27:35.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Deals","deals",247,"2026-09-25T15:26:22.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Hardware","hardware",183,"2026-09-25T13:41:27.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Science","science",139,"2026-09-25T11:55:23.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Consumer Tech","consumer-tech",132,"2026-09-25T15:30:00.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Software","software",88,"2026-09-24T23:06:55.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Dev Tools","dev-tools",81,"2026-09-25T09:59:40.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Startups","startups",73,"2026-09-25T14:05:04.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Gaming","gaming",47,"2026-09-25T13:33:16.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"General","general",46,"2026-09-25T02:12:57.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"Reviews","reviews",30,"2026-09-24T20:07:31.000Z",{"name":107,"slug":108,"count":109,"latest_published_at":110},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]