{"id":"https://openalex.org/W7164367896","doi":"https://doi.org/10.48550/arxiv.2606.12372","title":"UniIntervene: Agentic Intervention for Efficient Real-World Reinforcement Learning","display_name":"UniIntervene: Agentic Intervention for Efficient Real-World Reinforcement Learning","publication_year":2026,"publication_date":"2026-06-10","ids":{"openalex":"https://openalex.org/W7164367896","doi":"https://doi.org/10.48550/arxiv.2606.12372"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.12372","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12372","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.2606.12372","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138394779","display_name":"Haoyuan Deng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Haoyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138455804","display_name":"Yitong Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Yitong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136179834","display_name":"Yudong Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Yudong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138463893","display_name":"Haichao Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Haichao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138405293","display_name":"Zhenyu Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Zhenyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138413510","display_name":"Ziwei Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Ziwei","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":0.71670001745224,"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":0.71670001745224,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.06199999898672104,"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/T10653","display_name":"Robot Manipulation and Learning","score":0.02969999983906746,"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/intervention","display_name":"Intervention (counseling)","score":0.7396000027656555},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.7186999917030334},{"id":"https://openalex.org/keywords/psychological-intervention","display_name":"Psychological intervention","score":0.6700000166893005},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.6166999936103821},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.5788000226020813},{"id":"https://openalex.org/keywords/value","display_name":"Value (mathematics)","score":0.4681999981403351}],"concepts":[{"id":"https://openalex.org/C2780665704","wikidata":"https://www.wikidata.org/wiki/Q959298","display_name":"Intervention (counseling)","level":2,"score":0.7396000027656555},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7186999917030334},{"id":"https://openalex.org/C27415008","wikidata":"https://www.wikidata.org/wiki/Q7256382","display_name":"Psychological intervention","level":2,"score":0.6700000166893005},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.6166999936103821},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.5788000226020813},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.4993000030517578},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.4681999981403351},{"id":"https://openalex.org/C160145156","wikidata":"https://www.wikidata.org/wiki/Q778586","display_name":"Executable","level":2,"score":0.46470001339912415},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.46129998564720154},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.4410000145435333},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.42170000076293945},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.3718999922275543},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33660000562667847},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.2883000075817108}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.12372","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12372","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.2606.12372","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.12372","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":[{"score":0.7472124695777893,"display_name":"Decent work and economic growth","id":"https://metadata.un.org/sdg/8"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Human-in-the-loop":[0],"reinforcement":[1],"learning":[2],"(HiL-RL)":[3],"has":[4],"emerged":[5],"as":[6],"an":[7,55],"effective":[8],"paradigm":[9],"for":[10,172],"real-world":[11,47,174,180],"robotic":[12],"manipulation,":[13],"enabling":[14],"online":[15],"policy":[16,35,67],"improvement":[17],"with":[18],"human":[19,30,78,165,195],"guidance.":[20],"However,":[21],"current":[22,93],"HiL-RL":[23,202],"frameworks":[24],"remain":[25],"intervention-intensive,":[26],"relying":[27],"on":[28,108,178],"frequent":[29],"corrections":[31],"to":[32,200],"redirect":[33],"the":[34,66,73,88,92,122,187],"out":[36],"of":[37,75,91,143],"unproductive":[38,61],"exploration,":[39],"which":[40,100],"incurs":[41],"high":[42],"labor":[43],"cost":[44],"and":[45,63,95,118,147],"limits":[46],"scalability.":[48],"To":[49],"address":[50],"this,":[51,109],"we":[52],"propose":[53],"UniIntervene,":[54],"agentic":[56],"intervention":[57,120,131,145,162],"model":[58],"that":[59,184],"detects":[60],"exploration":[62],"autonomously":[64],"recovers":[65],"toward":[68],"high-value":[69,137],"states,":[70],"taking":[71],"over":[72],"bulk":[74],"interventions":[76,196],"from":[77,140,163],"operators.":[79],"Specifically,":[80],"UniIntervene":[81,134,160,185],"first":[82],"performs":[83],"future-conditioned":[84],"action-value":[85],"estimation,":[86],"predicting":[87],"latent":[89],"consequence":[90],"action":[94],"evaluating":[96],"its":[97],"induced":[98],"value,":[99],"provides":[101],"a":[102,110,136,141,153,168],"more":[103],"stable":[104],"progress":[105],"signal.":[106],"Building":[107],"temporal":[111],"value-risk":[112],"critic":[113],"aggregates":[114],"recent":[115],"value":[116,124],"dynamics":[117],"triggers":[119],"when":[121],"estimated":[123],"exhibits":[125],"sustained":[126],"stagnation":[127],"or":[128],"degradation.":[129],"When":[130],"is":[132],"required,":[133],"retrieves":[135],"recovery":[138,155,170],"target":[139],"memory":[142],"past":[144],"episodes":[146],"produces":[148],"executable":[149],"corrective":[150],"actions":[151],"through":[152],"goal-conditioned":[154],"policy.":[156],"In":[157],"this":[158],"way,":[159],"turns":[161],"passive":[164],"correction":[166],"into":[167],"value-aware":[169],"process":[171],"efficient":[173],"RL.":[175],"Extensive":[176],"experiments":[177],"diverse":[179],"manipulation":[181],"tasks":[182],"demonstrate":[183],"improves":[186],"average":[188],"success":[189],"rate":[190],"by":[191,197],"8.6%":[192],"while":[193],"reducing":[194],"57%":[198],"relative":[199],"state-of-the-art":[201],"baselines.":[203]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-12T00:00:00"}