{"id":"https://openalex.org/W7160925468","doi":"https://doi.org/10.48550/arxiv.2605.10821","title":"UniSteer: Unified Noise Steering for Efficient Human-Guided VLA Adaptation","display_name":"UniSteer: Unified Noise Steering for Efficient Human-Guided VLA Adaptation","publication_year":2026,"publication_date":"2026-05-11","ids":{"openalex":"https://openalex.org/W7160925468","doi":"https://doi.org/10.48550/arxiv.2605.10821"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.10821","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10821","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":null,"license_id":null,"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.2605.10821","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135939635","display_name":"Junjie Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Junjie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135946954","display_name":"Xinyao Qin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qin, Xinyao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135992840","display_name":"Yuhua Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Yuhua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135936357","display_name":"Kaixin Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Kaixin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080460228","display_name":"Chuheng Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Chuheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135988313","display_name":"Bin Liang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Bin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136000010","display_name":"Jun Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Jun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135935172","display_name":"Min Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Min","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135947317","display_name":"Li Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Li","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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.4528000056743622,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.4528000056743622,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/T10653","display_name":"Robot Manipulation and Learning","score":0.2320999950170517,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.17329999804496765,"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":0.7312999963760376},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.6743999719619751},{"id":"https://openalex.org/keywords/adaptation","display_name":"Adaptation (eye)","score":0.5902000069618225},{"id":"https://openalex.org/keywords/generator","display_name":"Generator (circuit theory)","score":0.5491999983787537},{"id":"https://openalex.org/keywords/noise-reduction","display_name":"Noise reduction","score":0.47450000047683716},{"id":"https://openalex.org/keywords/prior-probability","display_name":"Prior probability","score":0.43549999594688416}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7512999773025513},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7312999963760376},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.6743999719619751},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.5902000069618225},{"id":"https://openalex.org/C2780992000","wikidata":"https://www.wikidata.org/wiki/Q17016113","display_name":"Generator (circuit theory)","level":3,"score":0.5491999983787537},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.47450000047683716},{"id":"https://openalex.org/C177769412","wikidata":"https://www.wikidata.org/wiki/Q278090","display_name":"Prior probability","level":3,"score":0.43549999594688416},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4124000072479248},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.38850000500679016},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.30970001220703125},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3009999990463257},{"id":"https://openalex.org/C133731056","wikidata":"https://www.wikidata.org/wiki/Q4917288","display_name":"Control engineering","level":1,"score":0.2906999886035919},{"id":"https://openalex.org/C107418235","wikidata":"https://www.wikidata.org/wiki/Q1520565","display_name":"Human multitasking","level":2,"score":0.2599000036716461},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.2597000002861023}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.10821","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10821","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.10821","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10821","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":null,"license_id":null,"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":{"Diffusion-based":[0],"vision-language-action":[1],"(VLA)":[2],"models":[3],"have":[4],"emerged":[5],"as":[6,56],"strong":[7,174],"priors":[8],"for":[9,149],"robotic":[10],"manipulation,":[11],"yet":[12],"adapting":[13],"them":[14],"to":[15,77,140,187],"real-world":[16,43,196],"distributions":[17],"remains":[18],"challenging.":[19],"In":[20],"particular,":[21],"on-robot":[22],"reinforcement":[23,159],"learning":[24],"(RL)":[25],"is":[26,73,155],"expensive":[27],"and":[28,177],"time-consuming,":[29],"so":[30],"effective":[31],"adaptation":[32,197],"depends":[33],"on":[34,163,192],"efficient":[35],"policy":[36],"improvement":[37],"within":[38],"a":[39,57,63,112,130,142],"limited":[40,75],"budget":[41],"of":[42],"interactions.":[44],"Noise-space":[45],"RL":[46,124,176],"lowers":[47],"the":[48,52,68,136,150,182],"cost":[49],"by":[50],"keeping":[51],"pretrained":[53],"VLA":[54],"fixed":[55],"denoising":[58],"generator":[59],"while":[60],"updating":[61],"only":[62],"lightweight":[64],"actor":[65,153],"that":[66,117,154,168],"predicts":[67],"noise.":[69],"However,":[70],"its":[71],"performance":[72],"still":[74],"due":[76],"inefficient":[78],"autonomous":[79],"exploration.":[80],"Human":[81],"corrective":[82,120,132],"interventions":[83],"can":[84],"reduce":[85],"this":[86],"exploration":[87],"burden,":[88],"but":[89],"they":[90],"are":[91],"naturally":[92],"provided":[93],"in":[94,189],"action":[95],"space,":[96],"whereas":[97],"noise-space":[98,123,175],"finetuning":[99],"requires":[100],"supervision":[101],"over":[102],"noise":[103,143,152],"variables.":[104],"To":[105],"address":[106],"these":[107],"challenges,":[108],"we":[109],"propose":[110],"UniSteer,":[111],"Unified":[113],"Noise":[114],"Steering":[115],"framework":[116],"combines":[118],"human":[119,131],"guidance":[121,148],"with":[122],"through":[125],"approximate":[126],"action-to-noise":[127],"inversion.":[128],"Given":[129],"action,":[133],"UniSteer":[134,169],"inverts":[135],"frozen":[137],"flow-matching":[138],"decoder":[139],"recover":[141],"target,":[144],"which":[145],"provides":[146],"supervised":[147],"same":[151],"simultaneously":[156],"optimized":[157],"via":[158],"learning.":[160],"Real-world":[161],"experiments":[162],"diverse":[164],"manipulation":[165],"tasks":[166],"show":[167],"adapts":[170],"more":[171],"efficiently":[172],"than":[173],"action-space":[178],"human-in-the-loop":[179],"baselines,":[180],"improving":[181],"success":[183],"rate":[184],"from":[185],"20%":[186],"90%":[188],"66":[190],"minutes":[191],"average":[193],"across":[194],"four":[195],"tasks.":[198]},"counts_by_year":[],"updated_date":"2026-07-18T05:51:51.687321","created_date":"2026-05-13T00:00:00"}