arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.14945cs.ROcs.AI
收藏

TIDAL: 时间交错扩散与动作循环用于高频VLA控制

TIDAL: Temporally Interleaved Diffusion and Action Loop for High-Frequency VLA Control

Institute for Infocomm Research (I$^2$R), A*STAR, Singapore(信息通信研究所(I²R),A*STAR,新加坡);Tsinghua University, Beijing, China(清华大学,北京,中国);Nanyang Technological University, Singapore(南洋理工大学,新加坡)

Yuteng Sun, Haoran Wang, Ruofei Bai, Zhengguo Li, Jun Li, Meng Yee Michael Chuah, Wei Yun Yau

AI总结:

提出TIDAL分层框架,通过双频架构解耦语义推理与高频执行,实现边缘硬件上约9Hz控制更新,在动态拦截任务中性能提升2倍。

摘要:

大规模视觉-语言-动作(VLA)模型提供语义泛化能力,但推理延迟高,限制了其低频批处理执行范式。这种频率不匹配造成执行盲区,导致在目标于开环执行窗口内移动的动态环境中失败。我们提出TIDAL(时间交错扩散与动作循环),一种解耦语义推理与高频执行的分层框架。TIDAL作为扩散型VLA的骨干无关模块,采用双频架构重新分配计算预算。具体地,低频宏意图循环缓存语义嵌入,而高频微控制循环将单步流集成与执行交错进行。该设计在边缘硬件上实现约9Hz的控制更新(对比基线约2.4Hz),且不增加边际开销。为处理由此产生的延迟偏移,我们引入时间错位训练策略,策略利用过时语义意图与实时本体感觉学习预测性补偿。此外,我们通过差分运动预测器解决静态视觉编码器对速度不敏感的问题。TIDAL是架构性的,与系统级优化正交。实验表明,在动态拦截任务中,性能比开环基线提升2倍。尽管静态成功率略有下降,我们的方法将反馈频率提高4倍,并将语义嵌入的有效视野扩展到原生动作块大小之外。在非暂停推理协议下,TIDAL保持鲁棒,而标准基线因延迟失败。

英文摘要:

Large-scale Vision-Language-Action (VLA) models offer semantic generalization but suffer from high inference latency because they adopt a low-frequency batch-and-execute paradigm. This frequency mismatch creates an execution blind spot, causing failures in dynamic environments where targets move during the open-loop execution window. We propose TIDAL (Temporally Interleaved Diffusion and Action Loop), a hierarchical framework that decouples semantic reasoning from high-frequency actuation. TIDAL operates as a backbone-agnostic scheduler for diffusion-based VLAs, using a dual-frequency architecture to redistribute the computational budget. Specifically, a low-frequency macro-intent loop caches semantic embeddings, while a high-frequency micro-control loop interleaves single-step flow integration with execution, conditioning on the latest state fused with motion cues. To handle the resulting latency shift, we introduce a temporally misaligned training strategy where the policy learns stalenessaware compensation, conditioning on stale semantic intent alongside real-time proprioception. TIDAL is architectural, making it orthogonal to system-level optimizations. Experiments show an average 2.5x performance gain over open-loop baselines in dynamic interception tasks. Despite a slight decrease in static success rates, our approach yields a 4x increase in feedback frequency and extends the effective horizon of semantic embeddings beyond the native action chunk size. Under nonpaused physics and on a real robot, TIDAL remains robust to unmodeled dynamics, while standard open-loop baselines fail due to latency-induced error accumulation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑