Paper Detail
Qize Yu, Jiadi You, Yuran Wang, Jiaqi Liang, Bowen Ping, Yang Tian, Yue Chen, Minghong Cai, Zeying Gong, Ruihai Wu, Yinchuan Li, Junwei Liang, Yingcong Chen
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.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{yu2026affordancevla,
title = {AffordanceVLA: A Vision-Language-Action Model Empowering Action Generation through Affordance-Aware Understanding},
author = {Qize Yu and Jiadi You and Yuran Wang and Jiaqi Liang and Bowen Ping and Yang Tian and Yue Chen and Minghong Cai and Zeying Gong and Ruihai Wu and Yinchuan Li and Junwei Liang and Yingcong Chen},
year = {2026},
abstract = {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 e},
url = {https://huggingface.co/papers/2606.06155},
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, code available, huggingface daily},
eprint = {2606.06155},
archiveprefix = {arXiv},
}
{}