{"id":"https://openalex.org/W7161823710","doi":"https://doi.org/10.48550/arxiv.2605.19678","title":"RoVLA: Multi-Consistency Constraints for Robust Vision-Language-Action Models","display_name":"RoVLA: Multi-Consistency Constraints for Robust Vision-Language-Action Models","publication_year":2026,"publication_date":"2026-05-19","ids":{"openalex":"https://openalex.org/W7161823710","doi":"https://doi.org/10.48550/arxiv.2605.19678"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.19678","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.19678","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.2605.19678","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136563565","display_name":"Jingzhou Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Jingzhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136533899","display_name":"Yifan Wen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wen, Yifan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136380208","display_name":"Yongjie Bai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Yongjie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101289041","display_name":"Xinshuai Song","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Song, Xinshuai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136587064","display_name":"Yang Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136578860","display_name":"Liang Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Liang","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":0.6886000037193298,"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":0.6886000037193298,"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":0.08959999680519104,"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":0.06400000303983688,"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/robustness","display_name":"Robustness (evolution)","score":0.7620999813079834},{"id":"https://openalex.org/keywords/embodied-cognition","display_name":"Embodied cognition","score":0.42969998717308044},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.42309999465942383},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.33980000019073486},{"id":"https://openalex.org/keywords/task-analysis","display_name":"Task analysis","score":0.3352999985218048},{"id":"https://openalex.org/keywords/action-recognition","display_name":"Action recognition","score":0.29089999198913574}],"concepts":[{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.7620999813079834},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6930999755859375},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.47699999809265137},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4684999883174896},{"id":"https://openalex.org/C100609095","wikidata":"https://www.wikidata.org/wiki/Q1335050","display_name":"Embodied cognition","level":2,"score":0.42969998717308044},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.42309999465942383},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.33980000019073486},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3352999985218048},{"id":"https://openalex.org/C2987834672","wikidata":"https://www.wikidata.org/wiki/Q4677630","display_name":"Action recognition","level":3,"score":0.29089999198913574},{"id":"https://openalex.org/C48372109","wikidata":"https://www.wikidata.org/wiki/Q3913","display_name":"Binary number","level":2,"score":0.2718000113964081},{"id":"https://openalex.org/C23131810","wikidata":"https://www.wikidata.org/wiki/Q818574","display_name":"Observational study","level":2,"score":0.2614000141620636},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.25920000672340393},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.2549999952316284},{"id":"https://openalex.org/C175652121","wikidata":"https://www.wikidata.org/wiki/Q4379351","display_name":"Causal consistency","level":5,"score":0.25}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.19678","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.19678","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.2605.19678","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.19678","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":[{"display_name":"Quality Education","score":0.5947813987731934,"id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision-Language-Action":[0],"(VLA)":[1],"models":[2],"have":[3],"shown":[4],"strong":[5,180],"performance":[6],"on":[7,33,158,167],"embodied":[8,203],"manipulation,":[9],"yet":[10],"they":[11,67],"remain":[12],"brittle":[13],"under":[14,94,112,187],"visual":[15,135],"observation":[16,103,191],"changes,":[17],"paraphrased":[18],"language":[19],"instructions,":[20],"and":[21,50,102,128,136,144,161,164,171,183,190],"compounded":[22],"perturbations.":[23],"This":[24],"limitation":[25],"suggests":[26],"that":[27,176],"existing":[28],"methods":[29,182],"still":[30],"rely":[31],"heavily":[32],"shallow":[34],"correlations":[35,160],"in":[36],"the":[37,73,125,196],"training":[38],"distribution,":[39],"rather":[40],"than":[41],"learning":[42,200],"stable":[43,110],"couplings":[44],"among":[45],"task":[46,189],"semantics,":[47,99],"environment":[48],"states,":[49],"action":[51,122],"generation.":[52],"Although":[53],"recent":[54],"efforts":[55],"improve":[56],"robustness":[57,133,163,186],"through":[58],"larger-scale":[59],"training,":[60,154],"post-training":[61],"adaptation,":[62],"or":[63],"enhanced":[64],"predictive":[65],"modeling,":[66],"rarely":[68],"enforce":[69],"invariance-oriented":[70],"consistency":[71,93],"within":[72],"end-to-end":[74],"policy":[75],"itself.":[76],"To":[77],"address":[78],"this":[79],"issue,":[80],"we":[81],"propose":[82],"RoVLA,":[83],"a":[84],"robust":[85,202],"vision-language-action":[86],"framework":[87],"with":[88],"multi-consistency":[89,199],"constraints.":[90],"RoVLA":[91,155,177],"enforces":[92],"three":[95],"complementary":[96],"transformations:":[97],"instruction":[98,115],"trajectory":[100],"evolution,":[101],"perturbation.":[104],"Specifically,":[105],"Instructional":[106],"Consistency":[107,118,130],"(IC)":[108],"promotes":[109],"grounding":[111],"semantically":[113],"equivalent":[114],"rewrites,":[116],"Evolutionary":[117],"(EC)":[119],"preserves":[120],"coherent":[121],"intent":[123],"throughout":[124],"generation":[126],"process,":[127],"Observational":[129],"(OC)":[131],"improves":[132,162],"to":[134],"proprioceptive":[137],"perturbations":[138],"by":[139],"enforcing":[140],"consistent":[141],"predictions":[142],"before":[143],"after":[145],"targeted":[146],"disturbances.":[147],"By":[148],"explicitly":[149],"modeling":[150],"these":[151],"invariances":[152],"during":[153],"reduces":[156],"reliance":[157],"superficial":[159],"generalization.":[165],"Experiments":[166],"LIBERO-Plus,":[168],"RoboTwin":[169],"2.0,":[170],"real-world":[172],"manipulation":[173],"tasks":[174],"show":[175],"consistently":[178],"outperforms":[179],"baseline":[181],"exhibits":[184],"superior":[185],"diverse":[188],"shifts.":[192],"These":[193],"results":[194],"demonstrate":[195],"effectiveness":[197],"of":[198],"for":[201],"control.":[204],"Codes":[205],"will":[206],"be":[207],"available":[208],"at":[209],"https://github.com/HCPLab-SYSU/RoVLA.":[210]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-21T00:00:00"}