提出RIFT方法，让机器人动作模型无需逐帧生成未来就能高效决策。
Keep the Future, Drop the Rollout: RIFT for World Action Models
- 用预训练的未来令牌一次性构建未来状态缓存，跳过逐帧推演
- 在LIBERO任务上达到98.8%成功率，仅比原方法低0.2个百分点
- 适合需要低延迟部署的机器人系统，尤其适用于实时控制场景
世界动作模型（WAMs）依赖预测未来来决定机器人动作，但逐帧视频推演会增加部署延迟。我们探究动作生成是否必须依赖动态演化轨迹，还是仅需其未来表示。在40个LIBERO任务上，四类WAMs的闭环干预实验表明，遮蔽或重置未来缓存值会改变执行结果并降低成功率，说明模型对未来的值及其位置敏感。然而，对于Joint和Cosmos-2模型，复用一个固定的最终清洁键值（K/V）缓存几乎保持原始执行效果，末端执行器平均位移误差为1.7至1.9厘米，成功率维持在97.9%至98.2%。这揭示了缓存生成与消费的分离：模型可复用固定缓存，但仍需迭代推演构建它。为此，我们提出RIFT（Rollout-free Imagination via Future Tokens），通过学习的预测令牌在单次主干网络前向传播中构建完整未来K/V缓存，同时保留原有未来读取接口。在LIBERO上，RIFT实现98.8%成功率，接近基于推演的Joint、IDM和LingBot-VA（98.4%至98.6%），同时将动作块延迟降低68.2%至89.1%。在RoboTwin 2.0上，RIFT在干净/随机场景分别达到92.9%和92.6%的最高表现。结果表明，部署阶段可实现无推演的未来条件化，无需迭代视频生成。
原文摘要 · Abstract (English)
World action models (WAMs) condition robot actions on predicted futures, but iterative video rollout increases deployment latency. We ask whether action generation requires the evolving rollout trajectory or only its future representation. Across four WAMs on all 40 LIBERO tasks, paired closed-loop interventions show that masking or reassigning future-cache values changes execution and reduces success, indicating sensitivity to future values and their assigned positions. For Joint and Cosmos-2, however, replaying one fixed final-clean key/value (K/V) cache nearly preserves unmodified execution, with $1.7$ to $1.9$~cm end-effector average displacement error and $97.9\%$ to $98.2\%$ success. This separates cache consumption from production: these models can reuse a fixed cache but still require iterative rollout to construct it. We therefore propose RIFT (\emph{Rollout-free Imagination via Future Tokens}), which uses learned anticipation tokens to construct a complete future K/V cache in one backbone pass while retaining the original future-read interface. On LIBERO, RIFT achieves $98.8\%$ success, close to rollout-based Joint, IDM, and LingBot-VA at $98.4\%$ to $98.6\%$, while reducing action-chunk latency by $68.2\%$ to $89.1\%$. On RoboTwin~2.0, RIFT reaches $92.9/92.6\%$ on clean/randomized scenes, the highest observed among the evaluated methods. These results support rollout-free future conditioning without iterative video generation at deployment.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。