[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-model-learns-robot-actions-and-scene-changes-together":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},11049,"new-model-learns-robot-actions-and-scene-changes-together","New Model Learns Robot Actions and Scene Changes Together","One Transformer learns robot actions and future scene states together, matching simulation benchmarks and beating real-robot baselines on the Push-T task.","A new robotics model treats picking an action and predicting what happens next as the same language problem.\n\nResearchers built a system called Spatial Language Modeling that represents scene contours, goals, action targets, and future states with one shared vocabulary of coordinates and tokens. A single autoregressive Transformer learns to choose actions and to predict how the scene will change, both through the same next-token prediction objective. The team trained it from scratch: first on random, unstructured interactions with objects, then on expert demonstrations that pair action and state sequences. On the Push-T object-pushing task, the model posted competitive results in simulation and beat the real-robot policy baselines it was tested against on task success and target coverage.\n\nMost manipulation systems split \"what to do\" and \"what will happen\" into separate models or stages. Merging both into one next-token objective, using a single vocabulary of spatial tokens, is the actual contribution - a simpler training recipe that appears to hold up better on a physical robot. Ablations support that: joint action-and-state training beat action-only training, and pretraining on random play pushed performance up further.\n\nOne pushing task with one robot is nowhere near proof of a general-purpose robot brain, and \"competitive\" in simulation is not the same as \"winning\" - the real story here is better transfer to hardware, not a simulation breakthrough.","[\"robotics\",\"transformers\",\"manipulation research\",\"ai research\"]","2026-10-09T04:00:00.000Z","2026-10-10T03:45:23.771Z","2026-10-10T03:45:29.182Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Fix the dek: the source says the model only achieves 'competitive' (not superior) simulation performance and is higher-performing only versus the real-robot baselines, but the dek claims it is 'beating baselines on real and simulated tasks' — correct this overstatement so it matches the source.","resolved","ai",[32,33,34,35],"robotics","transformers","manipulation research","ai research",[37],{"name":38,"url":39},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.12172",0,{"sections":42},[43,46,50,55,60,64,68,73,78,83,87,92,97,102],{"name":44,"slug":30,"count":45,"latest_published_at":18},"AI",6783,{"name":47,"slug":48,"count":49,"latest_published_at":18},"Security","security",932,{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",486,"2026-10-08T22:40:11.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Deals","deals",474,"2026-10-08T22:00:00.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":18},"Hardware","hardware",232,{"name":65,"slug":66,"count":67,"latest_published_at":18},"Science","science",193,{"name":69,"slug":70,"count":71,"latest_published_at":72},"Consumer Tech","consumer-tech",181,"2026-10-08T23:26:35.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":77},"Startups","startups",117,"2026-10-08T16:45:00.000Z",{"name":79,"slug":80,"count":81,"latest_published_at":82},"Software","software",114,"2026-10-08T17:57:01.000Z",{"name":84,"slug":85,"count":86,"latest_published_at":18},"Dev Tools","dev-tools",106,{"name":88,"slug":89,"count":90,"latest_published_at":91},"General","general",66,"2026-10-09T04:46:11.000Z",{"name":93,"slug":94,"count":95,"latest_published_at":96},"Gaming","gaming",58,"2026-10-08T20:08:45.000Z",{"name":98,"slug":99,"count":100,"latest_published_at":101},"Reviews","reviews",34,"2026-10-08T14:00:22.000Z",{"name":103,"slug":104,"count":105,"latest_published_at":106},"How-To","how-to",8,"2026-10-05T09:00:00.000Z"]