{"id":"https://openalex.org/W7165647883","doi":"https://doi.org/10.48550/arxiv.2606.22540","title":"PolicyTrim: Boosting Intrinsic Policy Efficiency of Vision-Language-Action Models","display_name":"PolicyTrim: Boosting Intrinsic Policy Efficiency of Vision-Language-Action Models","publication_year":2026,"publication_date":"2026-06-21","ids":{"openalex":"https://openalex.org/W7165647883","doi":"https://doi.org/10.48550/arxiv.2606.22540"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.22540","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.22540","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.2606.22540","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139178340","display_name":"Xianghui Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xianghui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5071942112","display_name":"F Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Feng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139194793","display_name":"Wenbo Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Wenbo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123788085","display_name":"Hua Yan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan, Hua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139139248","display_name":"Zixuan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zixuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139132177","display_name":"Changsheng Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Changsheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5102831936","display_name":"Yinjie Lei","orcid":"https://orcid.org/0000-0001-6856-3342"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lei, Yinjie","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":0.7127000093460083,"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":0.7127000093460083,"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":0.05660000070929527,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.043699998408555984,"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/executable","display_name":"Executable","score":0.821399986743927},{"id":"https://openalex.org/keywords/boosting","display_name":"Boosting (machine learning)","score":0.7035999894142151},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.664900004863739},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6047000288963318},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6010000109672546},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5289000272750854},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.44920000433921814},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.44769999384880066}],"concepts":[{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.821399986743927},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8019999861717224},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.7035999894142151},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.664900004863739},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6047000288963318},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6010000109672546},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5289000272750854},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.44920000433921814},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.44769999384880066},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4293999969959259},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.39890000224113464},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38600000739097595},{"id":"https://openalex.org/C28761237","wikidata":"https://www.wikidata.org/wiki/Q7805321","display_name":"Time horizon","level":2,"score":0.3490999937057495},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.3273000121116638},{"id":"https://openalex.org/C166109690","wikidata":"https://www.wikidata.org/wiki/Q4677422","display_name":"Action selection","level":3,"score":0.3172000050544739},{"id":"https://openalex.org/C153701036","wikidata":"https://www.wikidata.org/wiki/Q659974","display_name":"Trustworthiness","level":2,"score":0.3068999946117401},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.29980000853538513},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.299699991941452},{"id":"https://openalex.org/C177212765","wikidata":"https://www.wikidata.org/wiki/Q627335","display_name":"Workflow","level":2,"score":0.2558000087738037},{"id":"https://openalex.org/C3018263672","wikidata":"https://www.wikidata.org/wiki/Q1296251","display_name":"Efficient algorithm","level":2,"score":0.2538999915122986}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.22540","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.22540","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.2606.22540","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.22540","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":[{"id":"https://metadata.un.org/sdg/16","display_name":"Peace, Justice and strong institutions","score":0.4610465466976166}],"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":[2,39,203],"provide":[3],"a":[4,68,122,145,174,226],"unified":[5],"paradigm":[6],"for":[7],"robotic":[8],"manipulation,":[9],"yet":[10],"their":[11],"real-world":[12],"deployment":[13,229],"is":[14,45],"often":[15],"bottlenecked":[16],"by":[17,48,211,218],"execution":[18,216],"efficiency.":[19],"While":[20],"existing":[21],"efforts":[22],"predominantly":[23],"focus":[24],"on":[25],"compute-centric":[26],"efficiency":[27,44],"to":[28,66,110,165,225],"reduce":[29],"per-step":[30],"inference":[31,80],"latency,":[32],"the":[33,52,61,75,103,129,152,161],"intrinsic":[34],"\\textbf{policy":[35],"efficiency}":[36],"of":[37,56,78,105,155],"these":[38],"remains":[40],"largely":[41],"unexplored.":[42],"Policy":[43],"fundamentally":[46],"affected":[47],"two":[49,71],"factors,":[50],"namely":[51],"effective":[53],"executable":[54,157],"length":[55,133],"predicted":[57],"action":[58,95,106,131,208],"chunks":[59,107],"and":[60,94,108,134,200,213],"total":[62,76],"physical":[63,114,137,193,215],"steps":[64,185,217],"required":[65],"complete":[67],"task.":[69],"These":[70],"factors":[72],"jointly":[73],"determine":[74],"number":[77],"forward":[79],"calls":[81],"during":[82],"execution.":[83],"We":[84],"observe":[85],"that":[86,127,149,177,205],"current":[87],"VLA":[88,202],"policies":[89],"struggle":[90],"with":[91,183],"planning":[92],"unreliability":[93],"redundancy,":[96],"suffering":[97],"from":[98],"severe":[99],"prediction":[100,163],"degradation":[101],"at":[102],"tail":[104],"tending":[109],"generate":[111],"unnecessarily":[112],"redundant":[113,136,192],"steps.":[115,138],"To":[116],"address":[117],"this,":[118],"we":[119,143,172],"propose":[120],"\\textbf{PolicyTrim},":[121],"reinforcement":[123],"learning-based":[124],"post-training":[125],"framework":[126,222],"extends":[128],"reliable":[130,140],"chunk":[132,141,209],"reduces":[135,214],"For":[139,169],"extension,":[142],"employ":[144],"dynamic":[146],"exploration":[147],"strategy":[148],"explicitly":[150],"rewards":[151],"successful":[153,180],"completion":[154],"longer":[156],"lengths,":[158],"progressively":[159],"pushing":[160],"trustworthy":[162],"horizon":[164],"its":[166],"empirical":[167],"limit.":[168],"step":[170],"efficiency,":[171],"design":[173],"redundancy-aware":[175],"reward":[176],"directly":[178],"favors":[179],"task":[181,233],"completions":[182],"fewer":[184],"while":[186],"penalizing":[187],"unreproducible":[188],"shortcuts,":[189],"effectively":[190],"eliminating":[191],"actions.":[194],"Extensive":[195],"experiments":[196],"across":[197],"three":[198,201],"benchmarks":[199],"demonstrate":[204],"PolicyTrim":[206],"improves":[207],"utilization":[210],"3$\\times$":[212],"51.4\\%.":[219],"Ultimately,":[220],"our":[221],"delivers":[223],"up":[224],"5.83$\\times$":[227],"end-to-end":[228],"speedup":[230],"without":[231],"compromising":[232],"success":[234],"rates.":[235]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-24T00:00:00"}