[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-robots-learn-better-when-you-stop-asking-yes-or-no":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},3127,"robots-learn-better-when-you-stop-asking-yes-or-no","Robots Learn Better When You Stop Asking Yes or No","A new preference-learning method lets human trainers define their own quality axes, boosting manipulation task success rates by 38 points over binary feedback.","Teaching robots what \"good\" looks like just got a lot more specific.\n\nResearchers have introduced Freeform Preference Learning (FPL), a training method that replaces the blunt yes\u002Fno of traditional preference labeling with something closer to how humans actually evaluate work. Instead of asking an annotator which robot movement was better overall, FPL lets them name the dimensions they care about — speed, safety, placement quality, carefulness — and rate pairs of trajectories along each axis independently. Those labels train a reward model that scores behavior per dimension, which in turn trains a policy that can optimize across all of them at once. Tested across four real-world and two simulated long-horizon manipulation tasks, FPL beat sparse-reward and binary-preference baselines by 38 percentage points.\n\nThe result matters because reward design has been one of robotics' most stubborn engineering taxes. Sparse success signals tell a robot almost nothing during the long middle of a task, and collapsing every quality trade-off into a single better\u002Fworse vote throws away information. FPL generates dense progress signals without requiring engineers to manually segment subtasks — a step that normally demands domain expertise and careful hand-tuning. The method also lets users steer behavior at test time by invoking different preference labels, without retraining.\n\nThe compositionality finding is the one to watch: the policy produced behaviors not present in the training data by combining learned axes. Whether that holds up at scale, or outside the lab's curated manipulation suite, is a question the paper doesn't answer yet — but it's the right question to be asking.","[\"robotics\",\"reinforcement-learning\",\"ai\",\"research\"]","2026-07-01T04:00:00.000Z","2026-07-01T07:51:07.411Z","2026-07-01T07:51:10.338Z","published",null,[],"ai",[26,27,24,28],"robotics","reinforcement-learning","research",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2606.32027",0,{"sections":35},[36,40,45,50,55,60,65,70,75,80,85,89,94,99],{"name":37,"slug":24,"count":38,"latest_published_at":39},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":41,"slug":42,"count":43,"latest_published_at":44},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":46,"slug":47,"count":48,"latest_published_at":49},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Dev Tools","dev-tools",59,"2026-07-07T04:00:00.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":86,"slug":87,"count":83,"latest_published_at":88},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"General","general",29,"2026-07-10T22:28:58.000Z",{"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"]