{"id":"https://openalex.org/W4416257571","doi":"https://doi.org/10.48550/arxiv.2509.18230","title":"Towards General Computer Control with Hierarchical Agents and Multi-Level Action Spaces","display_name":"Towards General Computer Control with Hierarchical Agents and Multi-Level Action Spaces","publication_year":2025,"publication_date":"2025-09-22","ids":{"openalex":"https://openalex.org/W4416257571","doi":"https://doi.org/10.48550/arxiv.2509.18230"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2509.18230","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.18230","pdf_url":"https://arxiv.org/pdf/2509.18230","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2509.18230","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Dong, Zihan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dong, Zihan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5009981236","display_name":"Xinyu Fan","orcid":"https://orcid.org/0000-0002-7626-4956","observed_orcids":["https://orcid.org/0000-0002-7626-4956"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fan, Xinyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Tang, Zixiang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Zixiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100656670","display_name":"Yunqing Li","orcid":"https://orcid.org/0000-0003-4460-9400","observed_orcids":["https://orcid.org/0000-0003-4460-9400"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yunqing","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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.7681000232696533,"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"}},"topics":[{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.7681000232696533,"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"}},{"id":"https://openalex.org/T10906","display_name":"AI-based Problem Solving and Planning","score":0.057999998331069946,"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"}},{"id":"https://openalex.org/T13904","display_name":"Artificial Intelligence Applications","score":0.013399999588727951,"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/hierarchical-reinforcement-learning","display_name":"hierarchical reinforcement learning","score":0.8149999976158142},{"id":"https://openalex.org/keywords/meta-actions","display_name":"meta-actions","score":0.6600000262260437},{"id":"https://openalex.org/keywords/early-stopping","display_name":"early stopping","score":0.4869999885559082},{"id":"https://openalex.org/keywords/on-device-inference","display_name":"on-device inference","score":0.7279999852180481},{"id":"https://openalex.org/keywords/inference-latency","display_name":"inference latency","score":0.48100000619888306}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8302000164985657},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6692000031471252},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.6198999881744385},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5885000228881836},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5855000019073486},{"id":"https://openalex.org/C115901376","wikidata":"https://www.wikidata.org/wiki/Q184199","display_name":"Automation","level":2,"score":0.47940000891685486},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.46889999508857727},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.46860000491142273},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4641000032424927},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4625999927520752},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.44999998807907104},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.4097000062465668},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4077000021934509},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.4072999954223633},{"id":"https://openalex.org/C144986985","wikidata":"https://www.wikidata.org/wiki/Q871236","display_name":"Hierarchical database model","level":2,"score":0.3935999870300293},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.37929999828338623},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.37459999322891235},{"id":"https://openalex.org/C124527596","wikidata":"https://www.wikidata.org/wiki/Q17029359","display_name":"Hierarchical control system","level":3,"score":0.36469998955726624},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.36410000920295715},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.335999995470047},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.31189998984336853},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.3100999891757965},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.3093000054359436},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.2939999997615814},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.27730000019073486},{"id":"https://openalex.org/C2779038628","wikidata":"https://www.wikidata.org/wiki/Q7248497","display_name":"Programming by demonstration","level":3,"score":0.2766000032424927},{"id":"https://openalex.org/C31170391","wikidata":"https://www.wikidata.org/wiki/Q188619","display_name":"Hierarchy","level":2,"score":0.26269999146461487},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.260699987411499},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.25450000166893005}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2509.18230","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.18230","pdf_url":"https://arxiv.org/pdf/2509.18230","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2509.18230","doi":"https://doi.org/10.48550/arxiv.2509.18230","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.18230","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"pmh:oai:arXiv.org:2509.18230","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.18230","pdf_url":"https://arxiv.org/pdf/2509.18230","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Controlling":[0],"desktop":[1,114],"applications":[2],"via":[3],"software":[4],"remains":[5],"a":[6,48,60,68,95,109,162],"fundamental":[7],"yet":[8],"under-served":[9],"problem.":[10],"Existing":[11],"multi-modal":[12],"large":[13],"language":[14],"models":[15],"(MLLMs)":[16],"ingest":[17],"screenshots":[18],"and":[19,25,42,65,80,93,125,151],"task":[20,73],"instructions":[21],"to":[22,77,89,166],"generate":[23],"keystrokes":[24],"mouse":[26],"events,":[27],"but":[28],"they":[29],"suffer":[30],"from":[31],"prohibitive":[32],"inference":[33,105,153],"latency,":[34],"poor":[35],"sample":[36],"efficiency":[37],"on":[38,120,127,138],"long-horizon":[39],"sparse-reward":[40],"tasks,":[41,115],"infeasible":[43],"on-device":[44,104],"deployment.":[45],"We":[46],"introduce":[47],"lightweight":[49],"hierarchical":[50,159],"reinforcement":[51],"learning":[52],"framework,":[53],"ComputerAgent,":[54],"that":[55,158],"formulates":[56],"OS":[57],"control":[58],"as":[59],"two-level":[61],"option":[62],"process":[63],"(manager":[64],"subpolicy),":[66],"employs":[67],"triple-modal":[69],"state":[70],"encoder":[71],"(screenshot,":[72],"ID,":[74],"numeric":[75],"state)":[76],"handle":[78],"visual":[79],"contextual":[81],"diversity,":[82],"integrates":[83],"meta-actions":[84],"with":[85],"an":[86],"early-stop":[87],"mechanism":[88],"reduce":[90],"wasted":[91],"interactions,":[92],"uses":[94],"compact":[96],"vision":[97],"backbone":[98],"plus":[99],"small":[100],"policy":[101],"networks":[102],"for":[103,170],"(15M":[106],"parameters).":[107],"On":[108],"suite":[110],"of":[111,149],"135":[112],"real-world":[113],"ComputerAgent":[116],"attains":[117],"92.1%":[118],"success":[119],"simple":[121,139],"tasks":[122,129],"(<8":[123],"steps)":[124],"58.8%":[126],"hard":[128],"(>=8":[130],"steps),":[131],"matching":[132],"or":[133],"exceeding":[134],"200B-parameter":[135],"MLLM":[136],"baselines":[137],"scenarios":[140],"while":[141],"reducing":[142],"model":[143],"size":[144],"by":[145],"over":[146],"four":[147],"orders":[148],"magnitude":[150],"halving":[152],"time.":[154],"These":[155],"results":[156],"demonstrate":[157],"RL":[160],"offers":[161],"practical,":[163],"scalable":[164],"alternative":[165],"monolithic":[167],"MLLM-based":[168],"automation":[169],"computer":[171],"control.":[172]},"counts_by_year":[],"updated_date":"2026-09-27T08:07:27.782135","created_date":"2025-10-10T00:00:00"}