{"paper":{"title":"Vision-Language Foundation Models as Effective Robot Imitators","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Simple fine-tuning adapts pre-trained vision-language models into robot policies that beat prior methods.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.RO","authors_text":"Chilam Cheang, Cunjun Yu, Hanbo Zhang, Hang Li, Hongtao Wu, Huaping Liu, Jie Xu, Minghuan Liu, Tao Kong, Weinan Zhang, Xinghang Li, Ya Jing","submitted_at":"2023-11-02T16:34:33Z","abstract_excerpt":"Recent progress in vision language foundation models has shown their ability to understand multimodal data and resolve complicated vision language tasks, including robotics manipulation. We seek a straightforward way of making use of existing vision-language models (VLMs) with simple fine-tuning on robotics data. To this end, we derive a simple and novel vision-language manipulation framework, dubbed RoboFlamingo, built upon the open-source VLMs, OpenFlamingo. Unlike prior works, RoboFlamingo utilizes pre-trained VLMs for single-step vision-language comprehension, models sequential history inf"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By exceeding the state-of-the-art performance with a large margin on the tested benchmark, we show RoboFlamingo can be an effective and competitive alternative to adapt VLMs to robot control.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That modest fine-tuning on existing language-conditioned manipulation datasets is sufficient to transfer the general vision-language understanding of pre-trained VLMs into reliable sequential robot policies without catastrophic forgetting or domain shift.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RoboFlamingo adapts open-source vision-language models for robot manipulation tasks via single-step comprehension plus an explicit policy head, outperforming prior methods on benchmarks with only light fine-tuning.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Simple fine-tuning adapts pre-trained vision-language models into robot policies that beat prior methods.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"91a85c646de6f8822037a18cc2a3b4a068bb69e07509d595515f345f2f48dba5"},"source":{"id":"2311.01378","kind":"arxiv","version":3},"verdict":{"id":"164d1108-8125-4063-9f30-d765cb47c30c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T21:40:52.339632Z","strongest_claim":"By exceeding the state-of-the-art performance with a large margin on the tested benchmark, we show RoboFlamingo can be an effective and competitive alternative to adapt VLMs to robot control.","one_line_summary":"RoboFlamingo adapts open-source vision-language models for robot manipulation tasks via single-step comprehension plus an explicit policy head, outperforming prior methods on benchmarks with only light fine-tuning.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That modest fine-tuning on existing language-conditioned manipulation datasets is sufficient to transfer the general vision-language understanding of pre-trained VLMs into reliable sequential robot policies without catastrophic forgetting or domain shift.","pith_extraction_headline":"Simple fine-tuning adapts pre-trained vision-language models into robot policies that beat prior methods."},"references":{"count":25,"sample":[{"doi":"","year":null,"title":"Do As I Can, Not As I Say: Grounding Language in Robotic Affordances","work_id":"037320f1-b0a9-4cbe-a639-bfb25409ce71","ref_index":1,"cited_arxiv_id":"2204.01691","is_internal_anchor":true},{"doi":"","year":null,"title":"OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Models","work_id":"87bfa84a-e663-4165-806f-93ef439d88d0","ref_index":2,"cited_arxiv_id":"2308.01390","is_internal_anchor":true},{"doi":"","year":null,"title":"S., Purohit, S., Reynolds, L., Tow, J., Wang, B., and Weinbach, S","work_id":"168a55d5-675d-49cf-be47-a17ee8cd742e","ref_index":3,"cited_arxiv_id":"2204.06745","is_internal_anchor":true},{"doi":"","year":null,"title":"RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control","work_id":"ff438a8a-8003-4fae-9131-acd418b3597b","ref_index":4,"cited_arxiv_id":"2307.15818","is_internal_anchor":true},{"doi":"","year":1901,"title":"Language models are few-shot learners","work_id":"677093e0-2019-45af-8c52-d9b33dec7e3d","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":25,"snapshot_sha256":"176287bd0ea2afcf183ff6e753d1f38baffb67f66ee0f0a9630a812af836f494","internal_anchors":13},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b2d9565cc7de3aef309fec141a2ba2e069e8709d5cb7b522e9343be85d23e6d7"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}