{"id":"https://openalex.org/W7157261871","doi":"https://doi.org/10.48550/arxiv.2604.24622","title":"CF-VLA: Efficient Coarse-to-Fine Action Generation for Vision-Language-Action Policies","display_name":"CF-VLA: Efficient Coarse-to-Fine Action Generation for Vision-Language-Action Policies","publication_year":2026,"publication_date":"2026-04-27","ids":{"openalex":"https://openalex.org/W7157261871","doi":"https://doi.org/10.48550/arxiv.2604.24622"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.24622","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.24622","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.2604.24622","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134766542","display_name":"Fan Du","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Du, Fan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134802758","display_name":"Feng Yan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan, Feng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134806225","display_name":"Jianxiong Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Jianxiong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134815970","display_name":"Xinrun Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Xinrun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134773386","display_name":"Weiye Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Weiye","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134786652","display_name":"Weinong Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Weinong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134767835","display_name":"Yu Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015115331","display_name":"Bin Qian","orcid":"https://orcid.org/0000-0002-0048-1487"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qian, Bin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134761390","display_name":"Zhihai He","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"He, Zhihai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wang, Fei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Fei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Yang, Heng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Heng","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7113999724388123,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.7113999724388123,"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/T10653","display_name":"Robot Manipulation and Learning","score":0.08060000091791153,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.0731000006198883,"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/initialization","display_name":"Initialization","score":0.6324999928474426},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5677000284194946},{"id":"https://openalex.org/keywords/residual","display_name":"Residual","score":0.5152999758720398},{"id":"https://openalex.org/keywords/thompson-sampling","display_name":"Thompson sampling","score":0.4787999987602234},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.4708999991416931},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.4661000072956085},{"id":"https://openalex.org/keywords/importance-sampling","display_name":"Importance sampling","score":0.4302000105381012},{"id":"https://openalex.org/keywords/boosting","display_name":"Boosting (machine learning)","score":0.41589999198913574},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.38940000534057617},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.3862000107765198}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6572999954223633},{"id":"https://openalex.org/C114466953","wikidata":"https://www.wikidata.org/wiki/Q6034165","display_name":"Initialization","level":2,"score":0.6324999928474426},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5677000284194946},{"id":"https://openalex.org/C155512373","wikidata":"https://www.wikidata.org/wiki/Q287450","display_name":"Residual","level":2,"score":0.5152999758720398},{"id":"https://openalex.org/C73602740","wikidata":"https://www.wikidata.org/wiki/Q7795822","display_name":"Thompson sampling","level":3,"score":0.4787999987602234},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4758000075817108},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.4708999991416931},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.4661000072956085},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.4302000105381012},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.41589999198913574},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3984000086784363},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.38940000534057617},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.3862000107765198},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.36480000615119934},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.3610999882221222},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.35569998621940613},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.334199994802475},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.334199994802475},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.32910001277923584},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.3197000026702881},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.3190000057220459},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.3181999921798706},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.3158000111579895},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3127000033855438},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.3107999861240387},{"id":"https://openalex.org/C175291020","wikidata":"https://www.wikidata.org/wiki/Q1156822","display_name":"Offset (computer science)","level":2,"score":0.29829999804496765},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.29789999127388},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.27959999442100525},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.27079999446868896},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.25760000944137573},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.2549000084400177},{"id":"https://openalex.org/C41426520","wikidata":"https://www.wikidata.org/wiki/Q1192065","display_name":"Point estimation","level":2,"score":0.25440001487731934}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.24622","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.24622","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.2604.24622","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.24622","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":[{"display_name":"No poverty","score":0.4855636954307556,"id":"https://metadata.un.org/sdg/1"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Flow-based":[0],"vision-language-action":[1],"(VLA)":[2],"policies":[3],"offer":[4],"strong":[5,152,213],"expressivity":[6],"for":[7],"action":[8,22,51,68,179],"generation,":[9],"but":[10],"suffer":[11],"from":[12,24,118],"a":[13,30,62,71,83,96,107,115,126,132,151],"fundamental":[14],"inefficiency:":[15],"multi-step":[16],"inference":[17],"is":[18,220],"required":[19],"to":[20,29,102],"recover":[21],"structure":[23],"uninformative":[25],"Gaussian":[26,104],"noise,":[27],"leading":[28],"poor":[31],"efficiency-quality":[32],"trade-off":[33],"under":[34,155],"real-time":[35],"constraints.":[36],"We":[37],"address":[38],"this":[39,119],"issue":[40],"by":[41,82,182,196,201],"rethinking":[42],"the":[43,46,56,92,111,171,186],"role":[44],"of":[45,54,158,192],"starting":[47,79],"point":[48],"in":[49],"generative":[50],"modeling.":[52],"Instead":[53],"shortening":[55],"sampling":[57,180],"trajectory,":[58],"we":[59,124],"propose":[60],"CF-VLA,":[61],"coarse-to-fine":[63,209],"two-stage":[64],"formulation":[65],"that":[66,75,87,129,147,207],"restructures":[67],"generation":[69,210],"into":[70,106],"coarse":[72,93,134],"initialization":[73],"step":[74],"constructs":[76],"an":[77],"action-aware":[78],"point,":[80],"followed":[81],"single-step":[84],"local":[85],"refinement":[86,117],"corrects":[88],"residual":[89],"errors.":[90],"Concretely,":[91],"stage":[94,113],"learns":[95,131],"conditional":[97],"posterior":[98],"over":[99],"endpoint":[100],"velocity":[101],"transform":[103],"noise":[105],"structured":[108],"initialization,":[109],"while":[110],"fine":[112],"performs":[114,138],"fixed-time":[116],"initialization.":[120],"To":[121],"stabilize":[122],"training,":[123],"introduce":[125],"stepwise":[127],"strategy":[128],"first":[130],"controlled":[133],"predictor":[135],"and":[136,144,184,199,215],"then":[137],"joint":[139],"optimization.":[140],"Experiments":[141],"on":[142,175],"CALVIN":[143],"LIBERO":[145],"show":[146],"our":[148],"method":[149],"establishes":[150],"efficiency-performance":[153],"frontier":[154],"low-NFE":[156],"(Number":[157],"Function":[159],"Evaluations)":[160],"regimes:":[161],"it":[162],"consistently":[163],"outperforms":[164],"existing":[165],"NFE=2":[166],"methods,":[167],"matches":[168],"or":[169],"surpasses":[170],"NFE=10":[172],"$\u03c0_{0.5}$":[173,200],"baseline":[174],"several":[176],"metrics,":[177],"reduces":[178],"latency":[181],"75.4%,":[183],"achieves":[185],"best":[187],"average":[188],"real-robot":[189],"success":[190],"rate":[191],"83.0%,":[193],"outperforming":[194],"MIP":[195],"19.5":[197],"points":[198],"4.0":[202],"points.":[203],"These":[204],"results":[205],"suggest":[206],"structured,":[208],"enables":[211],"both":[212],"performance":[214],"efficient":[216],"inference.":[217],"Our":[218],"code":[219],"available":[221],"at":[222],"https://github.com/EmbodiedAI-RoboTron/CF-VLA.":[223]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-29T00:00:00"}