# LY-GWM-RoboCasa-Human300：基于图动力学模型的预测引导动作选择

> 原文链接：https://robotics.hellobike.com/papers/2026-10-16-2610-23274
> 频道：papers | 更新日期：2026-10-16 | 标签：arxiv, cs.RO
> 核心摘要：该策略用图动力学预测与二值状态奖励从动作候选中选择，在RoboCasa任务本地和官方评测中验证。

# LY-GWM-RoboCasa-Human300：基于图动力学模型的预测引导动作选择

::: v-pre
> 原文链接：[LY-GWM-RoboCasa-Human300: Prediction-Guided Action Selection with a Graph Dynamics Model](https://arxiv.org/abs/2610.23274)（arXiv 2610.23274 · 发布日期：2026-10-16）

**LY-GWM-RoboCasa-Human300: Prediction-Guided Action Selection with a Graph Dynamics Model**

该策略用图动力学预测与二值状态奖励从动作候选中选择，在RoboCasa任务本地和官方评测中验证。

## 核心要点

- 动机是在推理时让未来状态预测影响发送到环境的动作，而非直接执行采样动作。
- 方法结合Human300微调的GR00T-N1.5动作生成器、动作条件图动力学模型和二值状态奖励评分器。
- 推理采样4个动作块，用16步预测时域评估奖励logit，执行预测奖励最高的候选。
- 本地成功率8.28%，历史基线7.24%；官方复评23.6%，论文描述公开复现入口。

**验证范围与局限**：验证限于RoboCasa任务：本地为RoboCasa 1.0.1预训练厨房划分50个任务，另有维护者官方复评50个任务。

## 英文摘要 (Abstract)

We present LY-GWM-RoboCasa-Human300, a policy that combines a Human300-fine-tuned GR00T-N1.5 action generator, an action-conditioned graph dynamics model, and a learned binary state-reward scorer. At inference, the policy samples four action chunks, predicts the future latent features and robot state for each, and executes the candidate with the highest predicted-state reward logit. Predictions therefore influence the actions sent to the environment. The evaluated system uses a 16-step prediction horizon and keeps all three components frozen during evaluation. In the local evaluation on 50 RoboCasa 1.0.1 tasks in the pretrain kitchen split, with 50 episodes per task, the system records 207/2,500 successes (8.28%), compared with 181/2,500 (7.24%) in the historical self-trained GR00T-N1.5 baseline. The observed difference is 26 successes, or 1.04 percentage points. The atomic group improves from 141/900 to 150/900. Separately, the RoboCasa maintainers re-evaluated the submission and reported 354/1,500 successes (23.6%) across 50 tasks, with leaderboard rates of 48.3%, 16.3%, and 3.1% for Atomic-Seen, Composite-Seen, and Composite-Unseen. The local and official results are reported separately; their difference is not attributed to a model improvement. Recorded local audits identify 212,700 changed environment-input chunks, including 28,223 in atomic tasks. We describe the actual training stages, implemented inference mechanism, evaluation provenance, and public reproduction entrypoints. These results document an active prediction-and-selection system and a positive historical difference.

作者：Tianliang Ma, Ting Ma, ZhongYuan Ma
:::

## 原始参考来源 (Sources)

- [arXiv 2610.23274：LY-GWM-RoboCasa-Human300: Prediction-Guided Action Selection with a Graph Dynamics Model](https://arxiv.org/abs/2610.23274)