[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-internw0-gives-robots-a-physical-world-model-for-lab-work":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},7561,"internw0-gives-robots-a-physical-world-model-for-lab-work","InternW0 Gives Robots a Physical World Model for Lab Work","Shanghai AI Laboratory's new model predicts video and controls robots at once, and it just ran real lab chemistry and pipetting tasks.","A new physical world model from Shanghai AI Laboratory predicts what a robot's cameras will see next and decides what its arm should do about it, in a single system.\n\nShanghai AI Laboratory has released InternW0, the first model in its InternW line of physical world models. The system pairs a heavyweight video-prediction expert with a lightweight, faster-running action expert, using a technique called flow matching to keep visual predictions and robot commands synchronized. Instead of recomputing its full future forecast every time a robot takes a new action, InternW0 reuses cached internal states and updates only the parts affected by new observations. The researchers trained it on roughly 7,200 hours of robot and first-person video data, including a 275-hour lab-specific dataset called EgoLab, then tested it on real science tasks: a 15-stage metal-organic framework synthesis and a 5-stage pipetting task that required force and tactile sensing.\n\nMost robot AI systems either predict video or issue motor commands efficiently, rarely both at once, which slows their reactions in labs where conditions keep changing. InternW0's asynchronous setup, a fast action loop paired with a slower video loop, is a direct answer to that lag. Testing it on real chemistry workflows instead of simple pick-and-place demos signals that Shanghai AI Laboratory is chasing lab automation, not warehouse robotics.\n\nThis is a research paper, not a shipping product, and the benchmarks are self-reported, so the real test is whether outside labs can reproduce the synthesis and pipetting results without Shanghai AI Laboratory's own equipment and training data.","[\"ai\",\"robotics\",\"world-models\",\"lab-automation\"]","2026-09-24T04:00:00.000Z","2026-09-24T06:46:30.173Z","2026-09-24T06:46:33.732Z","published",null,[],"ai",[24,26,27,28],"robotics","world-models","lab-automation",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.27656",0,{"sections":35},[36,39,43,48,53,58,63,68,73,78,83,88,93,98],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",4424,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",724,{"name":44,"slug":45,"count":46,"latest_published_at":47},"Policy","policy",380,"2026-09-23T22:53:43.000Z",{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",227,"2026-09-24T11:08:33.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":57},"Hardware","hardware",174,"2026-09-24T10:10:29.000Z",{"name":59,"slug":60,"count":61,"latest_published_at":62},"Science","science",136,"2026-09-24T09:00:00.000Z",{"name":64,"slug":65,"count":66,"latest_published_at":67},"Consumer Tech","consumer-tech",116,"2026-09-24T00:51:49.000Z",{"name":69,"slug":70,"count":71,"latest_published_at":72},"Software","software",85,"2026-09-23T20:00:00.000Z",{"name":74,"slug":75,"count":76,"latest_published_at":77},"Dev Tools","dev-tools",79,"2026-09-22T22:21:13.000Z",{"name":79,"slug":80,"count":81,"latest_published_at":82},"Startups","startups",66,"2026-09-23T17:28:38.000Z",{"name":84,"slug":85,"count":86,"latest_published_at":87},"Gaming","gaming",45,"2026-09-22T15:35:06.000Z",{"name":89,"slug":90,"count":91,"latest_published_at":92},"General","general",43,"2026-09-21T23:48:56.000Z",{"name":94,"slug":95,"count":96,"latest_published_at":97},"Reviews","reviews",27,"2026-09-22T13:00:00.000Z",{"name":99,"slug":100,"count":101,"latest_published_at":102},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]