外观
论文arXiv 2610.2136213 天前
LTM：面向 Vision-Language-Action 模型的基于潜在转移的上下文记忆
原文链接：LTM: Latent Transition Based In-Context Memory for Vision-Language-Action Models（arXiv 2610.21362 · 发布日期：2026-10-15）
LTM: Latent Transition Based In-Context Memory for Vision-Language-Action Models
针对操作中需记住不可见交互的问题，LTM以连续观测的潜在转移存储历史，并在RoboMME、LIBERO-Mem和真机计数上验证。
核心要点
- 动机：机器人操作常需记住已不在视野中的交互历史。
- 方法：以连续观测间潜在转移存历史，用冻结潜在动作编码器、轻量适配器和因果KV缓存。
- 结果：LTM-π在RoboMME达58.54%，在LIBERO-Mem达80.5%。
- 补充实验显示转移内容、延迟记忆准入、规模扩展、压缩和过程监督的作用。
验证范围与局限：验证范围包括RoboMME、LIBERO-Mem、Cup Swap诊断和真实机器人计数任务。
英文摘要 (Abstract)
Robotic manipulation often requires remembering interactions that are no longer visible. We introduce Latent Transition Memory (LTM), which stores history as compact latent transitions between consecutive observations. Instantiated on $π_{0.5}$ as LTM-$π$, our method uses a frozen latent action encoder, a lightweight adapter, and causal KV caching for incremental inference without memory-specific semantic annotations or external reasoning-model calls. With DreamDojo transitions, LTM-$π$ achieves 58.54% success on RoboMME and 80.5% on LIBERO-Mem, performing best among baselines without external reasoning models or memory-specific annotations. Ablations support the contribution of transition content beyond temporal cues and the efficiency of delayed memory admission. Scaling RoboMME demonstrations tenfold improves success from 57.67% to 87.50%, while a Cup Swap diagnostic highlights the role of process supervision in temporal composition. Under a 200-token memory budget, success remains 84.00%, versus 87.50% without compression. Finally, real-robot counting success improves from 11.25% to 82.50%. Together, these results demonstrate the effectiveness, scalability, and efficiency of latent transition memory, highlighting its capacity for complex temporal information, bounded-context operation, and real-world robotic control.
作者：Qixin Sun, WenLong Gao, Xiansheng Chen, Zhongyuan Wang, Si Liu, Pengwei Wang
来源
本页事实对应的公开页面，以原文为准。