{"id":"https://openalex.org/W7166868261","doi":"https://doi.org/10.18653/v1/2026.acl-long.1822","title":"On-policy Reinforcement Fine-tuning with Offline reward for Multi-step Embodied Planning","display_name":"On-policy Reinforcement Fine-tuning with Offline reward for Multi-step Embodied Planning","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166868261","doi":"https://doi.org/10.18653/v1/2026.acl-long.1822"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.acl-long.1822","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1822","pdf_url":"https://aclanthology.org/2026.acl-long.1822.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.acl-long.1822.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139852932","display_name":"Di Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Di Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139820338","display_name":"Jiaxin Fan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiaxin Fan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101259021","display_name":"Chloe Gu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chloe Gu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139770843","display_name":"Guanbo Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guanbo Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139843597","display_name":"Wei Yin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei Yin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139732776","display_name":"Wenhao Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenhao Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139759911","display_name":"Bo Jin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bo Jin","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.85724545,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"39277","last_page":"39307"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.5263000130653381,"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":0.5263000130653381,"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":0.2913999855518341,"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/T10586","display_name":"Robotic Path Planning Algorithms","score":0.036400001496076584,"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/reinforcement","display_name":"Reinforcement","score":0.4268999993801117},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.387800008058548},{"id":"https://openalex.org/keywords/embodied-cognition","display_name":"Embodied cognition","score":0.361299991607666},{"id":"https://openalex.org/keywords/action","display_name":"Action (physics)","score":0.3422999978065491},{"id":"https://openalex.org/keywords/control","display_name":"Control (management)","score":0.32359999418258667}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5178999900817871},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.4268999993801117},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4072999954223633},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.387800008058548},{"id":"https://openalex.org/C100609095","wikidata":"https://www.wikidata.org/wiki/Q1335050","display_name":"Embodied cognition","level":2,"score":0.361299991607666},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.3422999978065491},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.32359999418258667},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.30320000648498535},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.29409998655319214},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.27869999408721924},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.26330000162124634},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.26170000433921814}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.acl-long.1822","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1822","pdf_url":"https://aclanthology.org/2026.acl-long.1822.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.acl-long.1822","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.acl-long.1822","pdf_url":"https://aclanthology.org/2026.acl-long.1822.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6143107411","display_name":null,"funder_award_id":"62406270","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321885","display_name":"Science and Technology Commission of Shanghai Municipality","ror":"https://ror.org/03kt66j61"},{"id":"https://openalex.org/F4320327803","display_name":"Shanghai Rising-Star Program","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166868261.pdf","grobid_xml":"https://content.openalex.org/works/W7166868261.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Embodied":[0],"planning":[1],"requires":[2],"agents":[3],"to":[4,37,139,147],"make":[5],"coherent":[6],"multi-step":[7],"decisions":[8],"based":[9],"on":[10,96,118],"dynamic":[11],"visual":[12],"observations":[13],"and":[14,49,85,108,123,134,143,153],"verbal":[15],"goals.While":[16],"recent":[17,99],"vision-language":[18],"models":[19],"(VLMs)":[20],"excel":[21],"at":[22,155],"static":[23],"perception":[24],"tasks,":[25,104],"they":[26],"struggle":[27],"in":[28,52,131],"interactive":[29,102],"environments.Reinforcement":[30],"learning":[31],"(RL)":[32],"offers":[33],"a":[34,98],"natural":[35],"way":[36],"address":[38],"this":[39],"limitation,":[40],"yet":[41],"online":[42],"RL":[43],"approaches":[44],"suffer":[45],"from":[46],"costly":[47,83],"interaction":[48,84],"sparse":[50,86],"rewards":[51,66],"embodied":[53,103],"settings.This":[54],"paper":[55],"introduces":[56],"ORBIT,":[57],"an":[58],"On-policy":[59],"Reinforcement":[60],"finetuning":[61],"(RFT)":[62],"framework":[63],"with":[64],"offline":[65],"for":[67,101],"EmBodIed":[68],"Task":[69],"Planning,":[70],"that":[71,113],"preserves":[72],"the":[73,80],"generalization":[74,146],"benefits":[75],"of":[76,82],"RFT":[77],"while":[78,126],"addressing":[79],"challenges":[81],"rewards,":[87],"supported":[88],"by":[89],"solid":[90],"theoretical":[91],"guarantees.Our":[92],"approach":[93],"is":[94],"evaluated":[95],"EmbodiedBench,":[97],"benchmark":[100],"covering":[105],"both":[106],"in-domain":[107],"out-of-domain":[109],"scenarios.Experimental":[110],"results":[111],"show":[112],"ORBIT":[114],"achieves":[115],"SOTA":[116],"performance":[117],"EB-ALFRED,":[119],"outperforming":[120],"all":[121,151],"closed-source":[122],"online-RL-based":[124],"methods,":[125],"being":[127],"substantially":[128],"more":[129],"efficient":[130],"training":[132],"speed":[133],"computational":[135],"cost,":[136],"remaining":[137],"robust":[138],"sub-optimal":[140],"expert":[141],"trajectories,":[142],"exhibiting":[144],"strong":[145],"unseen":[148],"environments.We":[149],"released":[150],"code":[152],"data":[154],"https://github.com/mail-taii/Reinforced-Reasoning-for":[156]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}