{"id":"https://openalex.org/W7141392521","doi":"https://doi.org/10.48550/arxiv.2603.25685","title":"Persistent Robot World Models: Stabilizing Multi-Step Rollouts via Reinforcement Learning","display_name":"Persistent Robot World Models: Stabilizing Multi-Step Rollouts via Reinforcement Learning","publication_year":2026,"publication_date":"2026-03-26","ids":{"openalex":"https://openalex.org/W7141392521","doi":"https://doi.org/10.48550/arxiv.2603.25685"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.25685","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.25685","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.25685","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5130759814","display_name":"Jai Bardhan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bardhan, Jai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130756031","display_name":"Patrik Drozdik","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Drozdik, Patrik","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130808376","display_name":"Josef Sivic","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sivic, Josef","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5068303316","display_name":"Vladim\u00edr Petr\u00edk","orcid":"https://orcid.org/0000-0001-9450-4987"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Petrik, Vladimir","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.29030001163482666,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.29030001163482666,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10812","display_name":"Human Pose and Action Recognition","score":0.20509999990463257,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10531","display_name":"Advanced Vision and Imaging","score":0.1371999979019165,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.7258999943733215},{"id":"https://openalex.org/keywords/fidelity","display_name":"Fidelity","score":0.6816999912261963},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.6136999726295471},{"id":"https://openalex.org/keywords/robot","display_name":"Robot","score":0.5665000081062317},{"id":"https://openalex.org/keywords/convergence","display_name":"Convergence (economics)","score":0.392300009727478},{"id":"https://openalex.org/keywords/perception","display_name":"Perception","score":0.38909998536109924},{"id":"https://openalex.org/keywords/protocol","display_name":"Protocol (science)","score":0.3718000054359436},{"id":"https://openalex.org/keywords/robotics","display_name":"Robotics","score":0.3571999967098236},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.33649998903274536},{"id":"https://openalex.org/keywords/probabilistic-logic","display_name":"Probabilistic logic","score":0.3262999951839447}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7276999950408936},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7258999943733215},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.6816999912261963},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6194000244140625},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.6136999726295471},{"id":"https://openalex.org/C90509273","wikidata":"https://www.wikidata.org/wiki/Q11012","display_name":"Robot","level":2,"score":0.5665000081062317},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.392300009727478},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.38909998536109924},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3790999948978424},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.3718000054359436},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3646000027656555},{"id":"https://openalex.org/C34413123","wikidata":"https://www.wikidata.org/wiki/Q170978","display_name":"Robotics","level":3,"score":0.3571999967098236},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.33649998903274536},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.3262999951839447},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.3221000134944916},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.31859999895095825},{"id":"https://openalex.org/C60692881","wikidata":"https://www.wikidata.org/wiki/Q584529","display_name":"Humanoid robot","level":3,"score":0.3181999921798706},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.31540000438690186},{"id":"https://openalex.org/C113364801","wikidata":"https://www.wikidata.org/wiki/Q26674","display_name":"High fidelity","level":2,"score":0.31369999051094055},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.30979999899864197},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.303600013256073},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.30309998989105225},{"id":"https://openalex.org/C3261483","wikidata":"https://www.wikidata.org/wiki/Q119565","display_name":"Frame rate","level":2,"score":0.29490000009536743},{"id":"https://openalex.org/C166109690","wikidata":"https://www.wikidata.org/wiki/Q4677422","display_name":"Action selection","level":3,"score":0.29190000891685486},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.29030001163482666},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.2851000130176544},{"id":"https://openalex.org/C189645446","wikidata":"https://www.wikidata.org/wiki/Q350865","display_name":"Mirroring","level":2,"score":0.2786000072956085},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.2757999897003174},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.27230000495910645},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.2703999876976013},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.26420000195503235},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.263700008392334},{"id":"https://openalex.org/C171268870","wikidata":"https://www.wikidata.org/wiki/Q1486676","display_name":"GRASP","level":2,"score":0.25619998574256897},{"id":"https://openalex.org/C104114177","wikidata":"https://www.wikidata.org/wiki/Q79782","display_name":"Motion (physics)","level":2,"score":0.25279998779296875}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.25685","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.25685","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.25685","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.25685","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Action-conditioned":[0],"robot":[1,14],"world":[2,86],"models":[3,35,110],"generate":[4],"future":[5],"video":[6],"frames":[7],"of":[8,218],"the":[9,55,71,85,138,168,190,194,213],"manipulated":[10],"scene":[11],"given":[12],"a":[13,18,77,103,126,183,228],"action":[15],"sequence,":[16],"offering":[17],"promising":[19],"alternative":[20],"for":[21,38,54,108,171,186],"simulating":[22],"tasks":[23],"that":[24,83,116,129,156,179],"are":[25,36,165],"difficult":[26],"to":[27,59,64,111],"model":[28,87],"with":[29],"traditional":[30],"physics":[31],"engines.":[32],"However,":[33],"these":[34],"optimized":[37],"short-term":[39],"prediction":[40],"and":[41,61,114,131,164,221],"break":[42],"down":[43],"when":[44],"deployed":[45],"autoregressively:":[46],"each":[47],"predicted":[48],"clip":[49,169],"feeds":[50],"back":[51],"as":[52],"context":[53],"next,":[56],"causing":[57],"errors":[58],"compound":[60],"visual":[62,153],"quality":[63],"rapidly":[65],"degrade.":[66],"We":[67,98],"address":[68],"this":[69,100],"through":[70],"following":[72],"contributions.":[73],"First,":[74],"we":[75,124,149,177],"introduce":[76],"reinforcement":[78],"learning":[79],"(RL)":[80],"post-training":[81],"scheme":[82],"trains":[84],"on":[88,95,189,197,205,212],"its":[89,117],"own":[90],"autoregressive":[91],"rollouts":[92],"rather":[93],"than":[94],"ground-truth":[96],"histories.":[97],"achieve":[99],"by":[101,203,210],"adapting":[102],"recent":[104],"contrastive":[105],"RL":[106],"objective":[107],"diffusion":[109],"our":[112,180],"setting":[113],"show":[115,178],"convergence":[118],"guarantees":[119],"carry":[120],"over":[121,145],"exactly.":[122],"Second,":[123],"design":[125],"training":[127,174],"protocol":[128],"generates":[130],"compares":[132],"multiple":[133],"candidate":[134],"variable-length":[135],"futures":[136],"from":[137],"same":[139],"rollout":[140,187],"state,":[141],"reinforcing":[142],"higher-fidelity":[143],"predictions":[144],"lower-fidelity":[146],"ones.":[147],"Third,":[148],"develop":[150],"efficient,":[151],"multi-view":[152],"fidelity":[154,188],"rewards":[155],"combine":[157],"complementary":[158],"perceptual":[159],"metrics":[160,199],"across":[161],"camera":[162],"views":[163],"aggregated":[166],"at":[167],"level":[170],"dense,":[172],"low-variance":[173],"signal.":[175],"Fourth,":[176],"approach":[181],"establishes":[182],"new":[184],"state-of-the-art":[185],"DROID":[191],"dataset,":[192],"outperforming":[193],"strongest":[195],"baseline":[196],"all":[198],"(e.g.,":[200],"LPIPS":[201],"reduced":[202],"14%":[204],"external":[206],"cameras,":[207],"SSIM":[208],"improved":[209],"9.1%":[211],"wrist":[214],"camera),":[215],"winning":[216],"98%":[217],"paired":[219],"comparisons,":[220],"achieving":[222],"an":[223],"80%":[224],"preference":[225],"rate":[226],"in":[227],"blind":[229],"human":[230],"study.":[231]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-28T00:00:00"}