外观
论文arXiv 2609.284311 个月前
LiMA: 通过异步扩散将长期想象连接到实时灵巧操作
原文链接：LiMA: Bridging Long-term Imagination to Real-time Dexterous Manipulation via Asynchronous Diffusion（arXiv 2609.28431 · 发布日期：2026-09-23）
LiMA: Bridging Long-term Imagination to Real-time Dexterous Manipulation via Asynchronous Diffusion
LiMA通过异步双系统生成框架解耦意图规划与反应执行，在六个双手灵巧操作任务上验证降低延迟并提升成功率。
核心要点
- 动机是VLA缺乏细粒度动力学和空间感知，WAM迭代生成延迟高。
- LiMA以慢系统生成稀疏长时空意图，以快系统执行密集高频运动细化。
- Latent Schrödinger Bridge Coupling将细化表述为熵正则概率传输过程。
- 相较Cosmos-Policy推理延迟降低45.8%，总体成功率达70.8%，项目网站已公布。
验证范围与局限：验证范围为六个跨多种时域的双手灵巧操作任务及未见场景。
英文摘要 (Abstract)
Dexterous manipulation demands long-term foresight and rapid reactive control. Vision-Language-Action (VLA) models, while proficient in high-level reasoning, often lack a fine-grained understanding of physical dynamics and spatial perception. Conversely, World-Action Models (WAMs) typically suffer from high inference latency due to iterative generation. These deficiencies result in a critical temporal misalignment where the model's intent fails to adapt to rapid physical contact changes. To overcome this fundamental bottleneck, we propose LiMA, an asynchronous dual-system generative framework that systematically decouples intent planning from reactive execution. LiMA organizes computation into a multi-scale hierarchy: a slow system handles sparse long-horizon spatiotemporal intent generation, while a fast system focuses on dense high-frequency motion refinement. To align sparse intent predictions with dense action trajectories, we introduce a Latent Schrödinger Bridge Coupling mechanism that formulates refinement as an entropy-regularized probabilistic transport process. LiMA reduces inference latency by 45.8% compared with Cosmos-Policy via asynchronous decoupling. Evaluated across six bimanual dexterous manipulation tasks spanning multiple horizons, LiMA achieves an overall success rate of 70.8% and an average subtask success rate of 78.9%, while maintaining performance in unseen scenarios. The project website is available at https://ccdcs.github.io/LiMA_repo/
作者：Ning Chen, Yankai Fu, Junkai Zhao, Qianpu Sun, Guocai Yao, Pengwei Wang, Zhongyuan Wang, Shanghang Zhang
来源
本页事实对应的公开页面，以原文为准。