[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-robot-video-model-separates-scene-from-self":10,"sections":34},{"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":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},11131,"new-robot-video-model-separates-scene-from-self","New Robot Video Model Separates Scene From Self","A new benchmark and model split robot video prediction into background scenes and the robot's own actions to fix long-horizon task failures.","A new paper splits robot video prediction into two jobs: tracking the world and tracking the robot itself.\n\nResearchers describe a system called the World-Ego Model, which treats embodied video generation as two separate problems instead of one blended stream. The \"world\" covers the background and any objects the robot isn't currently touching, predicted mainly from visual history. The \"ego\" covers the robot and whatever it's manipulating, predicted mainly from the current instruction. The model pairs role-conditioned attention with a mixture-of-experts diffusion generator to keep those two predictions from interfering with each other. The team also built HTEWorld, a benchmark of 125,000 training clips and over 4.5 million frames covering combined navigation-and-manipulation tasks, plus 300 evaluation trajectories spanning more than 2,000 instructions.\n\nThe problem being solved is real: when one model predicts both a stable scene and precise instruction-following at once, it tends to get worse at both over long task sequences. That matters for any robot expected to navigate a space and then manipulate something in it, rather than just do one of those things in a lab demo. WEM reportedly beats prior approaches on the new benchmark while holding its own on existing manipulation-only tests.\n\nWorth noting: the headline benchmark is one the authors built themselves, so \"state of the art\" here means state of the art on their own yardstick. Still, separating world-modeling from self-modeling is a sensible architectural bet, and it's the kind of decomposition other embodied-AI labs will likely borrow.","[\"ai\",\"robotics\",\"video-generation\",\"world-models\"]","2026-10-09T04:00:00.000Z","2026-10-10T07:35:57.672Z","2026-10-10T07:36:03.481Z","published",null,[],"ai",[24,26,27,28],"robotics","video-generation","world-models",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.19957",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,90,95,100],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",6840,"2026-10-09T19:36:56.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",941,"2026-10-09T19:25:43.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",490,"2026-10-09T14:25:43.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",483,"2026-10-09T11:20:39.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",233,"2026-10-09T16:07:54.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Science","science",195,"2026-10-09T11:00:57.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Consumer Tech","consumer-tech",183,"2026-10-09T14:50:57.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Startups","startups",120,"2026-10-09T17:02:31.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Software","software",114,"2026-10-08T17:57:01.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Dev Tools","dev-tools",109,"2026-10-09T13:03:48.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"General","general",69,"2026-10-09T19:10:07.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"Gaming","gaming",59,"2026-10-09T11:43:43.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]