{"id":"https://openalex.org/W7135041823","doi":"https://doi.org/10.48550/arxiv.2603.10469","title":"DepthCache: Depth-Guided Training-Free Visual Token Merging for Vision-Language-Action Model Inference","display_name":"DepthCache: Depth-Guided Training-Free Visual Token Merging for Vision-Language-Action Model Inference","publication_year":2026,"publication_date":"2026-03-11","ids":{"openalex":"https://openalex.org/W7135041823","doi":"https://doi.org/10.48550/arxiv.2603.10469"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.10469","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.10469","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.2603.10469","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5069033450","display_name":"Yuquan Li","orcid":"https://orcid.org/0009-0007-0024-0732"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yuquan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128816802","display_name":"Lianjie Ma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Lianjie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128859690","display_name":"Han Ding","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ding, Han","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5070487709","display_name":"Lijun Zhu","orcid":"https://orcid.org/0000-0002-2655-1818"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Lijun","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":0.862500011920929,"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":0.862500011920929,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.03310000151395798,"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":0.018799999728798866,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6520000100135803},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.6391000151634216},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.6389999985694885},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5239999890327454},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.5001999735832214},{"id":"https://openalex.org/keywords/merge","display_name":"Merge (version control)","score":0.48809999227523804},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4020000100135803},{"id":"https://openalex.org/keywords/workspace","display_name":"Workspace","score":0.39890000224113464}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8299999833106995},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6520000100135803},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.6391000151634216},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.6389999985694885},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5442000031471252},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5239999890327454},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.5001999735832214},{"id":"https://openalex.org/C197129107","wikidata":"https://www.wikidata.org/wiki/Q1921621","display_name":"Merge (version control)","level":2,"score":0.48809999227523804},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4020000100135803},{"id":"https://openalex.org/C58581272","wikidata":"https://www.wikidata.org/wiki/Q12741163","display_name":"Workspace","level":3,"score":0.39890000224113464},{"id":"https://openalex.org/C205711294","wikidata":"https://www.wikidata.org/wiki/Q176953","display_name":"Rendering (computer graphics)","level":2,"score":0.39419999718666077},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.38040000200271606},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.3635999858379364},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.35440000891685486},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.35280001163482666},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.3418999910354614},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.32179999351501465},{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.31220000982284546},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.3003999888896942},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.2874000072479248},{"id":"https://openalex.org/C44210515","wikidata":"https://www.wikidata.org/wiki/Q16968978","display_name":"Bespoke","level":2,"score":0.27390000224113464},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2676999866962433},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.26570001244544983},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.2597000002861023},{"id":"https://openalex.org/C160234255","wikidata":"https://www.wikidata.org/wiki/Q812535","display_name":"Bayesian inference","level":3,"score":0.25839999318122864}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.10469","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.10469","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.2603.10469","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.10469","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":{"Vision-Language-Action":[0],"(VLA)":[1],"models":[2,146],"enable":[3],"generalist":[4],"robotic":[5,42],"manipulation":[6],"but":[7],"suffer":[8],"from":[9,16],"high":[10],"inference":[11,133],"latency.":[12],"This":[13],"bottleneck":[14],"stems":[15],"the":[17,37,74,79,88,125],"massive":[18],"number":[19],"of":[20],"visual":[21,58],"tokens":[22,34],"processed":[23],"by":[24],"large":[25],"language":[26],"backbones.":[27],"Existing":[28],"methods":[29],"either":[30],"prune":[31],"or":[32],"merge":[33,71],"uniformly,":[35],"degrading":[36],"spatial":[38],"reasoning":[39],"essential":[40],"for":[41,57],"control.":[43],"We":[44],"present":[45],"DepthCache,":[46],"a":[47,54,164],"training-free":[48],"framework":[49,114],"that":[50,168],"leverages":[51],"depth":[52],"as":[53],"structural":[55],"prior":[56],"token":[59],"compression.":[60,160],"It":[61],"partitions":[62],"observations":[63],"into":[64],"depth-based":[65],"regions":[66],"and":[67,152,174],"applies":[68],"spatially":[69],"differentiated":[70],"ratios,":[72],"preserving":[73],"near-field":[75],"workspace":[76],"while":[77,97],"compressing":[78],"distant":[80],"background.":[81],"To":[82],"exploit":[83],"temporal":[84],"redundancy,":[85],"DepthCache":[86,128,169],"distributes":[87],"merging":[89,153],"process":[90],"across":[91,120,143],"consecutive":[92],"frames,":[93],"ensuring":[94],"consistent":[95],"representations":[96],"reducing":[98],"per-step":[99],"computation.":[100],"A":[101],"motion-adaptive":[102],"pipeline":[103],"further":[104],"optimizes":[105],"auxiliary":[106],"view":[107],"compression":[108],"based":[109],"on":[110,163],"end-effector":[111],"dynamics.":[112],"The":[113],"requires":[115],"no":[116],"model":[117],"modification,":[118],"generalizing":[119],"diverse":[121],"VLA":[122,145],"architectures.":[123],"On":[124],"LIBERO":[126],"benchmark,":[127],"achieves":[129],"up":[130],"to":[131],"1.28x":[132],"speedup":[134],"with":[135],"less":[136],"than":[137],"1%":[138],"average":[139],"success":[140],"rate":[141],"degradation":[142,157],"three":[144],"(pi_0.5,":[147],"OpenVLA,":[148],"GR00T),":[149],"whereas":[150],"pruning":[151],"baselines":[154],"incur":[155],"4--24%":[156],"at":[158],"comparable":[159],"Real-world":[161],"experiments":[162],"physical":[165],"manipulator":[166],"demonstrate":[167],"enables":[170],"faster":[171],"task":[172],"throughput":[173],"more":[175],"responsive":[176],"closed-loop":[177],"control":[178],"in":[179],"latency-sensitive":[180],"scenarios.":[181]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-13T00:00:00"}