外观
论文HF Daily Papers · arXiv 2610.314236 天前
FocusVLA：意图引导动态视图路由提升操作成功率13.3%
本文由 AI 根据公开内容整理，可能有误，请以原文为准：FocusVLA: Future-Intent Guided Dynamic View Routing for Vision-Language-Action Models（HF Daily Papers · arXiv 2610.31423）
核心内容
现有视觉-语言-动作（VLA）模型在机器人操作中常过度依赖腕部视角，缺乏跨阶段动态调整全局与局部视图的能力，导致鲁棒性受限。研究团队提出FocusVLA阶段感知多视图路由框架，利用历史动作预测未来交互意图以指导视图选择。实验显示，该方法在RoboTwin 2.0基准上平均成功率较最强基线π_{0.5}提升10.10个百分点；真机评估中成功率从71.7%升至85.0%，实现13.3%的绝对增益。
FocusVLA的核心机制包含意图预测与动态路由模块，能根据任务阶段自适应协调全局环境观测与局部末端细节。该设计解决了多视图互补性利用不足的痛点，显著增强了策略在不同操作阶段的适应性。工程落地方面，这两个模块可作为插件集成至现有VLA框架，无需重构核心网络架构，为提升具身智能多模态感知效率提供了低成本改进路径。
关键看点
- 动态视图路由：基于历史动作预测未来交互意图，自适应切换全局与局部视图依赖，解决腕部视角过拟合问题
- 真机性能验证：在真实机器人操作评估中成功率由71.7%提升至85.0%，较基线π_{0.5}提高13.3个百分点
- 仿真基准突破：在RoboTwin 2.0数据集上平均成功率超越最强基线10.10pp，验证多视图互补性利用有效性
- 架构兼容设计：意图预测与路由模块支持即插即用，无需修改现有VLA核心架构即可完成集成升级
工程与产业观察
FocusVLA揭示了当前VLA模型在多视图融合上的盲区，证明了基于意图的动态感知调度比静态多模态输入更具工程价值。其插件式架构设计降低了算法迭代门槛，为具身智能从“被动看”向“主动聚焦”演进提供了可复用的技术范式，对提升复杂操作任务的泛化能力具有直接参考意义。
英文摘要 (Abstract)
Vision-language-action (VLA) models have become an important paradigm for robotic manipulation by unifying visual perception, language understanding, and action generation. Since reliable manipulation requires sufficient observation of objects, end-effectors, and their spatial relationships, multi-view inputs are widely used to provide complementary spatial information. However, our experiments show that existing VLA models do not fully exploit such multi-view complementarity. The policy tends to persistently rely on the wrist view, lacking the ability to dynamically adjust global-local view dependency across manipulation stages, which limits stage adaptability and robustness. To address this issue, we propose FocusVLA, a stage-aware multi-view routing framework. FocusVLA predicts future interaction intent from historical actions and uses it to guide dynamic view routing during action generation. By adaptively coordinating global and local observations across different task stages, FocusVLA improves multi-view utilization. On RoboTwin 2.0, FocusVLA improves the average success rate over the strongest baseline π_{0.5} by 10.10 pp. In real-world evaluation, FocusVLA improves the success rate from 71.7% to 85.0%, achieving a 13.3% gain over π_{0.5}. Moreover, the intent prediction and dynamic routing modules of FocusVLA can be integrated into existing VLA frameworks without modifying their core architectures.
Hugging Face Daily Papers 热度：1 upvotes 作者：xhx, Ziya Yu, Peng Li
来源
本页事实对应的公开页面，以原文为准。