RoboTTT: Context Scaling for Robot Policies

AK's Picks Embodied AI and Robotics Imitation Learning MLMRD ML TAVR Continual Learning
24
2026年07月16日
当前的机器人基础模型通常仅能处理单步或短历史的视觉-运动控制上下文。我们提出了“测试时训练机器人策略”（RoboTTT），这是一种新型机器人模型及其配套训练方法，可将视觉-运动控制上下文长度扩展至8000个时间步，较现有最先进策略提升了三个数量级，且不增加推理延迟。在如此长的上下文长度下，我们解锁了多项全新的机器人能力：仅需一段人类视频演示即可实现单次示例驱动的上下文内模仿学习；支持策略在运行过程中即时优化；对环境扰动具备更强鲁棒性；在多阶段、长时程任务中展现出更优性能。此外，我们首次观察到：随着预训练阶段所用上下文长度的增加，闭环控制性能呈现出持续稳定的提升趋势。RoboTTT的核心创新在于，将“测试时训练”（Test-Time Training）机制融入视觉-语言-动作（Vision-Language-Action）等机器人基础模型之中，构建出一种序列化模型——其循环状态由“快速权重”（fast weights）构成，即模型参数在训练与推理两个阶段均通过梯度下降动态更新，从而将历史信息高效压缩至权重空间，并据此检索上下文信息，实现长上下文条件建模。为实现训练上下文长度的规模化扩展，该训练方案融合了“序列动作强制”（sequence action forcing）与“截断式随时间反向传播”（truncated backpropagation through time）技术。在一系列具挑战性的真实机器人操作任务中，RoboTTT相较单步上下文基线模型的整体性能提升了87%，并首次成功完整执行了一项持续五分钟、包含十个阶段的装配任务——而所有基线模型均未能完成该任务。进一步对比表明，采用8000时间步上下文训练的RoboTTT，其性能比采用1000时间步上下文预训练的同一模型高出62%，这充分说明上下文长度可作为机器人基础模型的一项全新扩展维度。相关演示视频请访问：https://research.nvidia.com/labs/gear/robottt/
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/
许愿