[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-system-splits-robot-tasks-to-cut-long-horizon-failures":10,"sections":35},{"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":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},5500,"new-system-splits-robot-tasks-to-cut-long-horizon-failures","New System Splits Robot Tasks to Cut Long-Horizon Failures","BATON breaks multi-stage robot manipulation into separate subtasks with handoff tracking, lifting benchmark success by double digits over prior methods.","A new robotics framework tackles a stubborn problem: robots that can do a task's individual steps but still fail the sequence.\n\nResearchers describe BATON, a system for long-horizon robot manipulation, tasks that chain multiple contact-rich skills, like assembling an object step by step. Prior approaches paired a frozen vision-language-action (VLA) model with an LLM agent that plans in language and calls the VLA only for physical contact steps, but exploring a whole multi-stage task at test time gets expensive fast: a K-stage task with T episodes per stage needs roughly T^K attempts, and a failure does not reveal which stage broke it. BATON instead treats each subtask as its own unit of exploration, storing a solution for each in memory and composing the full trajectory afterward, which turns that multiplicative cost into an additive one. It also adds a transition-aware memory: a verifier agent checks the wrist camera view before calling the VLA, and handoff and lookahead steps make sure one subtask leaves the robot in a state the next subtask can actually use.\n\nThat second fix targets a quieter failure mode than error compounding: a subtask can succeed on its own terms while still leaving the robot in a position its successor cannot handle. On the RoboMemArena benchmark, BATON improved task success by 11.6 percent and cumulative success by 14.9 percent over the prior state of the art, without updating any model parameters.\n\nIt is a test-time fix, not a smarter robot: real progress on long-horizon manipulation will still need policies that don't need this much scaffolding to string a few actions together.","[\"robotics\",\"ai-agents\",\"long-horizon-planning\",\"benchmarks\"]","2026-08-18T04:00:00.000Z","2026-08-18T22:51:29.958Z","2026-08-18T22:51:41.859Z","published",null,[],"ai",[26,27,28,29],"robotics","ai-agents","long-horizon-planning","benchmarks",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2608.16889",0,{"sections":36},[37,41,45,50,55,60,65,70,75,79,84,89,94,99],{"name":38,"slug":24,"count":39,"latest_published_at":40},"AI",3293,"2026-08-20T04:00:00.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":40},"Security","security",435,{"name":46,"slug":47,"count":48,"latest_published_at":49},"Policy","policy",210,"2026-08-19T09:32:27.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",140,"2026-08-19T18:25:42.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",95,"2026-08-18T16:05:00.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Science","science",90,"2026-08-19T18:41:02.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Software","software",73,"2026-08-18T07:51:50.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":18},"Dev Tools","dev-tools",69,{"name":80,"slug":81,"count":82,"latest_published_at":83},"Startups","startups",47,"2026-08-19T19:13:46.000Z",{"name":85,"slug":86,"count":87,"latest_published_at":88},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"General","general",33,"2026-08-18T22:18:13.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]