{"id":"https://openalex.org/W7213529142","doi":"https://doi.org/10.48550/arxiv.2609.18207","title":"Reinforcement Learning for Real-Time Vision-Language-Action Policies","display_name":"Reinforcement Learning for Real-Time Vision-Language-Action Policies","publication_year":2026,"publication_date":"2026-09-16","ids":{"openalex":"https://openalex.org/W7213529142","doi":"https://doi.org/10.48550/arxiv.2609.18207"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.18207","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18207","pdf_url":"https://arxiv.org/pdf/2609.18207","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.18207","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5113246490","display_name":"Perry Dong","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dong, Perry","raw_affiliation_strings":["Stanford University https"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University https","institution_ids":["https://openalex.org/I97018004"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079506491","display_name":"Kuo-Han Hung","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hung, Kuo-Han","raw_affiliation_strings":["Stanford University https"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University https","institution_ids":["https://openalex.org/I97018004"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080725225","display_name":"Dorsa Sadigh","orcid":"https://orcid.org/0000-0002-7802-9183","observed_orcids":["https://orcid.org/0000-0002-7802-9183"]},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Sadigh, Dorsa","raw_affiliation_strings":["Stanford University https"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University https","institution_ids":["https://openalex.org/I97018004"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5005431772","display_name":"Chelsea Finn","orcid":"https://orcid.org/0000-0001-6298-0874","observed_orcids":["https://orcid.org/0000-0001-6298-0874"]},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Finn, Chelsea","raw_affiliation_strings":["Stanford University https"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Stanford University https","institution_ids":["https://openalex.org/I97018004"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I97018004"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":5,"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":0.95169997215271,"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":0.95169997215271,"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":0.047200001776218414,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.00019999999494757503,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/vision-language-action-models","display_name":"vision-language-action models","score":0.9279999732971191},{"id":"https://openalex.org/keywords/reinforcement-learning-fine-tuning","display_name":"reinforcement learning fine-tuning","score":0.8289999961853027},{"id":"https://openalex.org/keywords/real-time-robot-control","display_name":"real-time robot control","score":0.7950000166893005},{"id":"https://openalex.org/keywords/inference-latency","display_name":"inference latency","score":0.9259999990463257},{"id":"https://openalex.org/keywords/distribution-shift","display_name":"distribution shift","score":0.7929999828338623},{"id":"https://openalex.org/keywords/action-chunking","display_name":"action chunking","score":0.7710000276565552},{"id":"https://openalex.org/keywords/robotic-manipulation","display_name":"robotic manipulation","score":0.6470000147819519},{"id":"https://openalex.org/keywords/online-adaptation","display_name":"online adaptation","score":0.3449999988079071},{"id":"https://openalex.org/keywords/sample-efficiency","display_name":"sample efficiency","score":0.5600000023841858},{"id":"https://openalex.org/keywords/ball-balancing","display_name":"ball balancing","score":0.8960000276565552}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.8686000108718872},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7946000099182129},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6039000153541565},{"id":"https://openalex.org/C90509273","wikidata":"https://www.wikidata.org/wiki/Q11012","display_name":"Robot","level":2,"score":0.4846999943256378},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.4643000066280365},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.4569999873638153},{"id":"https://openalex.org/C166109690","wikidata":"https://www.wikidata.org/wiki/Q4677422","display_name":"Action selection","level":3,"score":0.4198000133037567},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.4163999855518341},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.4106999933719635},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4025999903678894},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4023999869823456},{"id":"https://openalex.org/C199190896","wikidata":"https://www.wikidata.org/wiki/Q3509276","display_name":"Learning classifier system","level":3,"score":0.37389999628067017},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.3702999949455261},{"id":"https://openalex.org/C2776544517","wikidata":"https://www.wikidata.org/wiki/Q189447","display_name":"Unexpected events","level":2,"score":0.3400000035762787},{"id":"https://openalex.org/C45235069","wikidata":"https://www.wikidata.org/wiki/Q278425","display_name":"Table (database)","level":2,"score":0.33079999685287476},{"id":"https://openalex.org/C188888258","wikidata":"https://www.wikidata.org/wiki/Q7353390","display_name":"Robot learning","level":4,"score":0.3142000138759613},{"id":"https://openalex.org/C188116033","wikidata":"https://www.wikidata.org/wiki/Q2664563","display_name":"Q-learning","level":3,"score":0.27320000529289246},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2700999975204468},{"id":"https://openalex.org/C77967617","wikidata":"https://www.wikidata.org/wiki/Q4677561","display_name":"Active learning (machine learning)","level":2,"score":0.26179999113082886},{"id":"https://openalex.org/C196340769","wikidata":"https://www.wikidata.org/wiki/Q7698910","display_name":"Temporal difference learning","level":3,"score":0.25060001015663147}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.18207","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18207","pdf_url":"https://arxiv.org/pdf/2609.18207","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2609.18207","doi":"https://doi.org/10.48550/arxiv.2609.18207","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.18207","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.18207","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.18207","pdf_url":"https://arxiv.org/pdf/2609.18207","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G6022427645","display_name":"CAREER: Safe and Influencing Interactions for Human-Robot Systems","funder_award_id":"1941722","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7213529142.pdf","grobid_xml":"https://content.openalex.org/works/W7213529142.grobid-xml"},"referenced_works_count":28,"referenced_works":["https://openalex.org/W2575705757","https://openalex.org/W2781726626","https://openalex.org/W2904246096","https://openalex.org/W2964161785","https://openalex.org/W2968340082","https://openalex.org/W2986616324","https://openalex.org/W4323076570","https://openalex.org/W4402354047","https://openalex.org/W4405254492","https://openalex.org/W4413362511","https://openalex.org/W4414079054","https://openalex.org/W4414079285","https://openalex.org/W4414581628","https://openalex.org/W4414968801","https://openalex.org/W4415251629","https://openalex.org/W4415332297","https://openalex.org/W4416398156","https://openalex.org/W4416958104","https://openalex.org/W7103139230","https://openalex.org/W7103171635","https://openalex.org/W7127366176","https://openalex.org/W7139920986","https://openalex.org/W7155175472","https://openalex.org/W7157012058","https://openalex.org/W7162393363","https://openalex.org/W7170084358","https://openalex.org/W7171669197","https://openalex.org/W7171815610"],"related_works":["https://openalex.org/W7162393363","https://openalex.org/W7157012058","https://openalex.org/W4414079285","https://openalex.org/W7123834068","https://openalex.org/W7131476132","https://openalex.org/W4416157228","https://openalex.org/W4414050921","https://openalex.org/W4416255006","https://openalex.org/W7164810662","https://openalex.org/W7222308244"],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1,77],"fine-tuning":[2,98,120],"on":[3,75,112,166],"top":[4],"of":[5,20,67,105,207],"large,":[6],"pretrained":[7,137],"Vision-Language-Action":[8],"(VLA)":[9],"models":[10,25],"offers":[11],"promise":[12],"for":[13,82,116,179],"highly":[14],"reliable":[15,118],"robot":[16,215,229],"deployment.":[17],"However,":[18],"because":[19],"their":[21],"scale,":[22],"modern":[23],"VLA":[24,119,138,182],"suffer":[26],"from":[27,130,241],"high":[28],"inference":[29],"latency,":[30,68],"so":[31],"the":[32,65,85,101,167,185,196],"observation":[33],"used":[34],"to":[35,63,161,194,243,253],"select":[36],"an":[37,176],"action":[38,128,133,140],"is":[39],"often":[40],"stale":[41],"by":[42,95,156],"execution":[43,62],"time,":[44],"creating":[45],"a":[46,114,135,148,191],"distribution":[47,87],"shift":[48],"that":[49,99],"can":[50],"substantially":[51],"degrade":[52],"reliability":[53],"and":[54,78,124,201,223],"performance.":[55],"Prior":[56],"work":[57],"has":[58],"explored":[59],"asynchronous":[60],"policy":[61,151,193,239],"reduce":[64],"effect":[66],"but":[69],"these":[70],"methods":[71,203],"are":[72],"mostly":[73],"built":[74],"imitation":[76],"offer":[79],"no":[80],"mechanism":[81],"moving":[83],"beyond":[84],"training":[86],"toward":[88],"higher":[89],"reliability.":[90],"We":[91,170],"close":[92],"this":[93,172],"gap":[94],"enabling":[96],"RL":[97,177],"meets":[100],"real-time":[102,181],"control":[103],"requirements":[104],"dynamic":[106,212,224],"real-world":[107,213,255],"manipulation.":[108],"Our":[109],"approach":[110],"builds":[111],"EXPO-FT,":[113,175],"framework":[115,178],"sample-efficient,":[117],"with":[121,227],"reinforcement":[122],"learning,":[123],"decouples":[125],"slow,":[126],"expressive":[127],"generation":[129],"fast,":[131,153],"reactive":[132,154],"edits:":[134],"large":[136],"proposes":[139],"chunks":[141],"using":[142],"its":[143],"strong":[144],"behavior":[145],"prior,":[146],"while":[147],"lightweight":[149],"edit":[150],"performs":[152],"decision-making":[155],"editing":[157],"actions":[158],"in":[159,163,204],"response":[160],"changes":[162],"state,":[164],"conditioned":[165],"latest":[168],"observation.":[169],"instantiate":[171],"as":[173],"Real-Time":[174,188,235],"finetuning":[180],"policies.":[183],"On":[184,210],"Kinetix":[186],"benchmark,":[187],"EXPO-FT":[189,236],"enables":[190],"delayed":[192,200],"achieve":[195],"best":[197],"performance":[198,240],"among":[199],"non-delayed":[202],"10":[205,208,233],"out":[206],"environments.":[209],"four":[211],"tasks,":[214],"object":[216,225],"passing,":[217],"ball":[218],"balancing,":[219],"table":[220],"soccer":[221],"kicking,":[222],"picking,":[226],"online":[228],"data":[230],"capped":[231],"at":[232],"minutes,":[234],"improves":[237],"average":[238],"42%":[242],"97%,":[244],"all":[245],"without":[246],"human":[247],"intervention,":[248],"demonstrating":[249],"rapid,":[250],"sample-efficient":[251],"adaptation":[252],"challenging":[254],"dynamics.":[256],"Website:":[257],"https://pd-perry.github.io/real-time-expo-ft":[258]},"counts_by_year":[{"year":2026,"cited_by_count":5}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-18T00:00:00"}