{"id":"https://openalex.org/W7156937340","doi":"https://doi.org/10.48550/arxiv.2604.24447","title":"Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot Deployment","display_name":"Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot Deployment","publication_year":2026,"publication_date":"2026-04-27","ids":{"openalex":"https://openalex.org/W7156937340","doi":"https://doi.org/10.48550/arxiv.2604.24447"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.24447","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.24447","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.24447","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134766897","display_name":"Kaijun Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Kaijun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134798666","display_name":"Qiwei Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Qiwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134807988","display_name":"Da Peng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peng, Da","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134758646","display_name":"Zhiyang Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zhiyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134758505","display_name":"Xijun Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Xijun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134781897","display_name":"Jinyu Gu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gu, Jinyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.3806000053882599,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.3806000053882599,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.18160000443458557,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic 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/T10191","display_name":"Robotics and Sensor-Based Localization","score":0.06679999828338623,"subfield":{"id":"https://openalex.org/subfields/2202","display_name":"Aerospace 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/software-deployment","display_name":"Software deployment","score":0.7979000210762024},{"id":"https://openalex.org/keywords/asynchronous-communication","display_name":"Asynchronous communication","score":0.6556000113487244},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.6481999754905701},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5806000232696533},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.48739999532699585},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.48579999804496765},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.4853000044822693},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.4309999942779541}],"concepts":[{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.7979000210762024},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6985999941825867},{"id":"https://openalex.org/C151319957","wikidata":"https://www.wikidata.org/wiki/Q752739","display_name":"Asynchronous communication","level":2,"score":0.6556000113487244},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.6481999754905701},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5806000232696533},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.48840001225471497},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.48739999532699585},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.48579999804496765},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.4853000044822693},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.4309999942779541},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.4242999851703644},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.37049999833106995},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.36489999294281006},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.36059999465942383},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.35929998755455017},{"id":"https://openalex.org/C2777087702","wikidata":"https://www.wikidata.org/wiki/Q7306421","display_name":"Redux","level":2,"score":0.3490999937057495},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.33469998836517334},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.32690000534057617},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.32350000739097595},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.32190001010894775},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.31209999322891235},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.2996000051498413},{"id":"https://openalex.org/C90509273","wikidata":"https://www.wikidata.org/wiki/Q11012","display_name":"Robot","level":2,"score":0.2971000075340271},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.2939000129699707},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2761000096797943},{"id":"https://openalex.org/C34413123","wikidata":"https://www.wikidata.org/wiki/Q170978","display_name":"Robotics","level":3,"score":0.27000001072883606},{"id":"https://openalex.org/C2780378061","wikidata":"https://www.wikidata.org/wiki/Q25351891","display_name":"Service (business)","level":2,"score":0.2621000111103058},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.2565999925136566},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.25459998846054077}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.24447","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.24447","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.24447","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.24447","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":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","score":0.4223250150680542,"display_name":"Affordable and clean energy"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision-Language-Action":[0],"(VLA)":[1],"models":[2],"are":[3],"promising":[4],"for":[5,46],"generalist":[6],"robot":[7],"control,":[8],"but":[9],"on-robot":[10],"deployment":[11,49],"is":[12,152],"bottlenecked":[13],"by":[14,36,100,114],"real-time":[15],"inference":[16,93],"under":[17,63],"tight":[18],"cost":[19],"and":[20,33,59,109,120,126,138],"energy":[21],"budgets.":[22],"Most":[23],"prior":[24],"evaluations":[25],"rely":[26],"on":[27,136,140],"desktop-grade":[28],"GPUs,":[29],"obscuring":[30],"the":[31],"trade-offs":[32],"opportunities":[34],"offered":[35],"heterogeneous":[37],"edge":[38,71,141],"accelerators":[39],"(GPUs/XPUs/NPUs).":[40],"We":[41],"present":[42],"a":[43,56,90,95,101],"systematic":[44],"analysis":[45],"low-cost":[47],"VLA":[48],"via":[50],"model-hardware":[51,61],"co-characterization.":[52],"First,":[53],"we":[54,88,117],"build":[55],"cross-accelerator":[57],"leaderboard":[58,150],"evaluate":[60],"pairs":[62],"CET":[64],"(Cost,":[65],"Energy,":[66],"Time),":[67],"showing":[68],"that":[69],"right-sized":[70],"devices":[72],"can":[73],"be":[74],"more":[75],"cost-/energy-efficient":[76],"than":[77],"flagship":[78],"GPUs":[79,137],"while":[80],"meeting":[81],"control-rate":[82],"constraints.":[83],"Second,":[84],"using":[85],"in-depth":[86],"profiling,":[87],"uncover":[89],"consistent":[91],"two-phase":[92],"pattern:":[94],"compute-bound":[96],"VLM":[97],"backbone":[98],"followed":[99],"memory-bound":[102],"Action":[103],"Expert,":[104],"which":[105],"induces":[106],"phase-dependent":[107],"underutilization":[108],"hardware":[110],"inefficiency.":[111],"Finally,":[112],"guided":[113],"these":[115],"insights,":[116],"propose":[118],"DP-Cache":[119],"V-AEFusion":[121],"to":[122,133],"reduce":[123],"diffusion":[124],"redundancy":[125],"enable":[127],"asynchronous":[128],"pipeline":[129],"parallelism,":[130],"achieving":[131],"up":[132],"2.9x":[134],"speedup":[135],"6x":[139],"NPUs":[142],"with":[143],"only":[144],"marginal":[145],"success":[146],"degradation.":[147],"The":[148],"example":[149],"website":[151],"available":[153],"at:":[154],"https://vla-leaderboard-01.vercel.app/.":[155]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-04-29T00:00:00"}