AffordanceVLA introduces a structured affordance-forecasting bridge for VLA models, enabling robots to reason about what to manipulate, where to interact, and how to act for more robust instruction-following manipulation.</p>\n","updatedAt":"2026-06-05T15:41:05.465Z","author":{"_id":"692fa5d17ff1da99eb783dfb","avatarUrl":"/avatars/5477343d26250cbda7babb8f1fdee49d.svg","fullname":"Qize Yu","name":"Skywalker0410","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8529626131057739},"editors":["Skywalker0410"],"editorAvatarUrls":["/avatars/5477343d26250cbda7babb8f1fdee49d.svg"],"reactions":[{"reaction":"🤗","users":["Skywalker0410","hellouniverse"],"count":2}],"isReport":false}},{"id":"6a22f1391b95e49c2fa18fe9","author":{"_id":"65144605be453924e0519d9d","avatarUrl":"/avatars/763446333a2270abaedfdb26041370cb.svg","fullname":"huang","name":"hellouniverse","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false},"createdAt":"2026-06-05T15:54:33.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"Excellent work with plenty of insightful takeaways.","html":"<p>Excellent work with plenty of insightful takeaways.</p>\n","updatedAt":"2026-06-05T15:54:33.085Z","author":{"_id":"65144605be453924e0519d9d","avatarUrl":"/avatars/763446333a2270abaedfdb26041370cb.svg","fullname":"huang","name":"hellouniverse","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.8757949471473694},"editors":["hellouniverse"],"editorAvatarUrls":["/avatars/763446333a2270abaedfdb26041370cb.svg"],"reactions":[{"reaction":"❤️","users":["Skywalker0410"],"count":1}],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2606.06155","authors":[{"_id":"6a22426b3490a593e87b14f2","user":{"_id":"692fa5d17ff1da99eb783dfb","avatarUrl":"/avatars/5477343d26250cbda7babb8f1fdee49d.svg","isPro":false,"fullname":"Qize Yu","user":"Skywalker0410","type":"user","name":"Skywalker0410"},"name":"Qize Yu","status":"claimed_verified","statusLastChangedAt":"2026-06-05T15:06:41.260Z","hidden":false},{"_id":"6a22426b3490a593e87b14f3","name":"Jiadi You","hidden":false},{"_id":"6a22426b3490a593e87b14f4","name":"Yuran Wang","hidden":false},{"_id":"6a22426b3490a593e87b14f5","name":"Jiaqi Liang","hidden":false},{"_id":"6a22426b3490a593e87b14f6","name":"Bowen Ping","hidden":false},{"_id":"6a22426b3490a593e87b14f7","name":"Yang Tian","hidden":false},{"_id":"6a22426b3490a593e87b14f8","name":"Yue Chen","hidden":false},{"_id":"6a22426b3490a593e87b14f9","name":"Minghong Cai","hidden":false},{"_id":"6a22426b3490a593e87b14fa","name":"Zeying Gong","hidden":false},{"_id":"6a22426b3490a593e87b14fb","name":"Ruihai Wu","hidden":false},{"_id":"6a22426b3490a593e87b14fc","name":"Yinchuan Li","hidden":false},{"_id":"6a22426b3490a593e87b14fd","name":"Junwei Liang","hidden":false},{"_id":"6a22426b3490a593e87b14fe","name":"Yingcong Chen","hidden":false}],"publishedAt":"2026-06-04T00:00:00.000Z","submittedOnDailyAt":"2026-06-05T00:00:00.000Z","title":"AffordanceVLA: A Vision-Language-Action Model Empowering Action Generation through Affordance-Aware Understanding","submittedOnDailyBy":{"_id":"692fa5d17ff1da99eb783dfb","avatarUrl":"/avatars/5477343d26250cbda7babb8f1fdee49d.svg","isPro":false,"fullname":"Qize Yu","user":"Skywalker0410","type":"user","name":"Skywalker0410"},"summary":"Vision-Language-Action (VLA) models leverage the rich world knowledge of pretrained vision-language models (VLMs) to enable instruction-following robotic manipulation. However, the structural mismatch between VLM semantic spaces and embodied control policies often hinders the learning of precise perception--action mappings. To address this challenge, we propose AffordanceVLA, a unified framework that introduces structured affordance forecasting as a task-oriented intermediate representation to establish a more precise and robust perception--action mapping. Specifically, we progressively model manipulation priors through three complementary components: 1) Which2Act for object-centric grounding via visual latent prediction to suppress distractions; 2) Where2Act for 2D interaction localization via affordance map estimation; and 3) How2Act for 3D geometric reasoning to guide manipulation policies. These affordance cues provide spatially grounded, semantically conditioned, and action-coupled intermediate representations, thereby naturally bridging vision, language and action. We integrate these modules into a Mixture-of-Transformer (MoT) architecture with specialized experts and train the model using a three-stage training strategy with a progressive data curriculum. To overcome the scarcity of dense affordance labels in robotic datasets, we also develop a robust automated data augmentation pipeline. Extensive experiments on simulation and real-world demonstrate that AffordanceVLA achieves strong performance across diverse manipulation scenarios.","upvotes":3,"discussionId":"6a22426c3490a593e87b14ff","projectPage":"https://skywalker-yqz.github.io/AffordanceVLA/","githubRepo":"https://github.com/Skywalker-yqz/AffordanceVLA","githubRepoAddedBy":"user","ai_summary":"AffordanceVLA introduces a unified framework that uses structured affordance forecasting as an intermediate representation to improve the precision of perception-action mapping in robotic manipulation by leveraging vision-language models.","ai_keywords":["Vision-Language-Action models","vision-language models","embodied control policies","affordance forecasting","visual latent prediction","affordance map estimation","3D geometric reasoning","Mixture-of-Transformer","automated data augmentation"],"ai_summary_model":"Qwen/Qwen2.5-Coder-32B-Instruct","githubStars":1,"organization":{"_id":"61dcd8e344f59573371b5cb6","name":"PekingUniversity","fullname":"Peking University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/vavgrBsnkSejriUF4lXDE.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"692fa5d17ff1da99eb783dfb","avatarUrl":"/avatars/5477343d26250cbda7babb8f1fdee49d.svg","isPro":false,"fullname":"Qize Yu","user":"Skywalker0410","type":"user"},{"_id":"6844057801bb8ad58ca2bc17","avatarUrl":"/avatars/a4eb908a3d3bfdd424ea74e5a93aadf7.svg","isPro":false,"fullname":"Nimol","user":"Nimolty","type":"user"},{"_id":"6744b49365b98acef35a2e02","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/7jF2GSdosob8i--tRBxSQ.png","isPro":false,"fullname":"chen-boyu","user":"chen-by","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"61dcd8e344f59573371b5cb6","name":"PekingUniversity","fullname":"Peking University","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/vavgrBsnkSejriUF4lXDE.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2606/2606.06155.md"}">
AffordanceVLA: A Vision-Language-Action Model Empowering Action Generation through Affordance-Aware Understanding
Authors: ,
,
,
,
,
,
,
,
,
,
,
Abstract
AffordanceVLA introduces a unified framework that uses structured affordance forecasting as an intermediate representation to improve the precision of perception-action mapping in robotic manipulation by leveraging vision-language models.
Vision-Language-Action (VLA) models leverage the rich world knowledge of pretrained vision-language models (VLMs) to enable instruction-following robotic manipulation. However, the structural mismatch between VLM semantic spaces and embodied control policies often hinders the learning of precise perception--action mappings. To address this challenge, we propose AffordanceVLA, a unified framework that introduces structured affordance forecasting as a task-oriented intermediate representation to establish a more precise and robust perception--action mapping. Specifically, we progressively model manipulation priors through three complementary components: 1) Which2Act for object-centric grounding via visual latent prediction to suppress distractions; 2) Where2Act for 2D interaction localization via affordance map estimation; and 3) How2Act for 3D geometric reasoning to guide manipulation policies. These affordance cues provide spatially grounded, semantically conditioned, and action-coupled intermediate representations, thereby naturally bridging vision, language and action. We integrate these modules into a Mixture-of-Transformer (MoT) architecture with specialized experts and train the model using a three-stage training strategy with a progressive data curriculum. To overcome the scarcity of dense affordance labels in robotic datasets, we also develop a robust automated data augmentation pipeline. Extensive experiments on simulation and real-world demonstrate that AffordanceVLA achieves strong performance across diverse manipulation scenarios.
Community
AffordanceVLA introduces a structured affordance-forecasting bridge for VLA models, enabling robots to reason about what to manipulate, where to interact, and how to act for more robust instruction-following manipulation.
Excellent work with plenty of insightful takeaways.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2606.06155 in a model README.md to link it from this page.
Cite arxiv.org/abs/2606.06155 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2606.06155 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.