# <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(238, 249, 253);">(2026) Afford-VLA: Action-Aligned Visual Planning via Internalized Affordance</span></span>
| <!-- --> |
|---|
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(219, 238, 221);">作者:</span></span><span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(219, 238, 221);"> Runze Wang; Yuqian Fu; Yu Li; Tao Lin; Tianwen Qian; Mohamed Elhoseiny; Bo Zhao; Yanwei Fu; Yu-Gang Jiang; Xiangyang Xue;</span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">期刊: </span></span><span style="color: rgb(255, 0, 0);"><span style="background-color: rgb(243, 250, 244);">, </span></span><span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">2026.</span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(219, 238, 221);">期刊分区:</span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">本地链接: </span></span><span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);"><a href="zotero://open-pdf/0_RBZ9L22W" rel="noopener noreferrer nofollow">Wang 等 - 2026 - Afford-VLA Action-Aligned Visual Planning via Internalized Affordance.pdf</a></span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(219, 238, 221);">DOI: </span></span><span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(219, 238, 221);"><a href="https://doi.org/10.48550/ARXIV.2605.24203" rel="noopener noreferrer nofollow">10.48550/ARXIV.2605.24203</a></span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">摘要: </span></span><span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">Vision-language-action (VLA) models have shown strong potential for generalist robot manipulation, yet they remain limited by insufficient spatial reasoning, particularly in determining where to interact in complex visual scenes. While recent efforts introduce various forms of visual planning to address this issue, existing approaches either rely on global geometric cues, symbolic intermediate representations, or externally generated visual signals, which are often weakly coupled with downstream action prediction. In this work, we revisit visual planning in VLA systems and argue that effective planning should be local, visually grounded, internally generated, and directly aligned with action. Based on this insight, we propose Afford-VLA, a unified framework that internalizes task-conditioned affordance as an explicit visual planning interface within VLA models. Concretely, we introduce learnable tokens to query task-relevant interaction regions, decode affordance masks from multimodal features, and convert them into compact embeddings that directly condition action generation. This design enables affordance to be both generated and utilized within the VLA, forming a tightly coupled perception–action pathway. To further support this integration, we adopt a training strategy that allows the affordance pathway to be jointly optimized with action prediction, improving its effectiveness for downstream control. We evaluate our method on multiple simulation benchmarks, including LIBERO, LIBERO-Plus, and SimplerEnv, achieving consistent state-of-the-art performance, along with strong real-world results. These findings demonstrate that internalizing affordance as action-aligned visual planning provides a powerful paradigm for improving VLA systems. Codes and Models will be released at Afford-VLA.</span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(219, 238, 221);">标签:</span></span> |
| <span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">笔记日期: </span></span><span style="color: rgb(25, 60, 71);"><span style="background-color: rgb(243, 250, 244);">2026/6/4 13:01:45</span></span> |
# <span style="color: rgb(224, 255, 255);"><span style="background-color: rgb(102, 205, 170);">📜 研究核心</span></span>
Tips: 做了什么，解决了什么问题，创新点与不足？
# ⚙️ 内容
# 💡 创新点
# 🧩 不足
# <span style="color: rgb(32, 178, 170);"><span style="background-color: rgb(175, 238, 238);">🔁 研究内容</span></span>
# 💧 数据
# 👩🏻💻 方法
# 🔬 实验
# 📜 结论
# <span style="color: rgb(0, 77, 153);"><span style="background-color: rgb(135, 206, 250);">🤔 个人总结</span></span>
Tips: 你对哪些内容产生了疑问，你认为可以如何改进？
