---
title: "RoboTTT：通过上下文缩放实现机器人策略的测试时训练"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.15275v1"
permalink: "https://feed.astralor.com/items/itm_745c16e5b4b25039b35424fd"
published: "2026-07-16T17:59:06.000Z"
score: 79
channel: "ai"
tags: ["NVIDIA","RoboTTT","Vision-Language-Action","上下文缩放","测试时训练"]
---

# RoboTTT：通过上下文缩放实现机器人策略的测试时训练

## AI 点评

> 这篇论文值得关注是因为它证明了机器人基础模型中也存在‘上下文缩放’规律，类似大语言模型的科学直觉。对于机器人研究人员和从业者，RoboTTT提供了一条在无需重新训练整个模型的情况下通过更长上下文提升性能的路径。尤其对于从事长时任务规划和机器人操作的公司，这项技术可能显著缩短部署时间。

## 速览

论文提出RoboTTT，一种将机器人视觉-运动上下文扩展到8000时间步的新方法，比现有最优策略提升三个数量级且不增加推理延迟。RoboTTT集成测试时训练到视觉-语言-动作基础模型中，实现从人类视频演示的一次性上下文模仿、在线策略改进、对扰动的鲁棒性，并在多阶段长时任务中表现优异。在真实机器人操作任务上，相比单步基线性能提升87%，完整完成5分钟10阶段组装任务。8000步上下文预训练比1000步性能提升62%，表明上下文长度可作为机器人基础模型的新缩放维度。

## 关键点

- RoboTTT将视觉-运动上下文扩展到8000时间步，比现有最优策略高三个数量级。
- 集成测试时训练，通过梯度更新实现超长上下文压缩与检索，不增加推理延迟。
- 在真实机器人任务上性能提升87%，完整完成5分钟10阶段复杂组装任务。
- 表明上下文长度是机器人基础模型的新缩放维度，8000步比1000步预训练提升62%。

## 全文

Recent robot foundation models operate with single-step or short-history visuomotor context. We introduce Test-Time-Training Robot Policies (RoboTTT), a robot model and training recipe that scale visuomotor context to 8K timesteps, three orders of magnitude beyond state-of-the-art policies, without growing inference latency. At this context length, we unlock new robot capabilities: one-shot in-context imitation from human video demonstrations, on-the-fly policy improvement, robustness to perturbations, and stronger performance on multi-stage, long-horizon tasks. We also observe, for the first time, steady gains in closed-loop performance as pretraining context length scales. At its core, RoboTTT integrates Test-Time Training into robot foundation models such as Vision-Language-Action policies, yielding a sequence model whose recurrent state consists of fast weights, parameters updated by gradient descent during both training and inference, compressing histories into weight space and retrieving contextual information for long-context conditioning. To scale training context length, the recipe combines sequence action forcing with truncated backpropagation through time. On challenging real-robot manipulation tasks, RoboTTT improves overall performance by 87% over the single-step context baseline and fully completes a five-minute, ten-stage assembly task, which no baseline ever does. RoboTTT trained with 8K-timestep context outperforms the same model pretrained with 1K timesteps by 62%, suggesting context length as a new scaling axis for robot foundation models. Videos are available at https://research.nvidia.com/labs/gear/robottt/

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_745c16e5b4b25039b35424fd