arXiv 2410.200182024-10-26cs.ROcs.AIcs.CVcs.LG

GHIL-Glue：使用过滤后的子目标图像进行分层控制

GHIL-Glue: Hierarchical Control with Filtered Subgoal Images

Toyota Research Institute（丰田研究所）；UC Berkeley（加州大学伯克利分校）；Princeton University（普林斯顿大学）

Kyle B. Hatch, Ashwin Balakrishna, Oier Mees, Suraj Nair, Seohong Park, Blake Wulfe, Masha Itkina, Benjamin Eysenbach, Sergey Levine, Thomas Kollar, Benjamin Burchfiel

备注：: Code, model checkpoints and videos can be found at https://ghil-glue.github.io

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出 GHIL-Glue，通过过滤无任务进展的子目标并增强策略对视觉伪影的鲁棒性，连接生成式预测模型与低层控制策略，在 CALVIN 及真实零样本操控任务中取得显著提升。

摘要：

在互联网规模数据上预训练的图像和视频生成模型，可以极大提升机器人学习系统的泛化能力。这些模型能够充当高层规划器，生成中间子目标，供低层的目标条件策略去达成。然而，这类系统的性能可能会受到生成模型与低层控制器之间接口的严重制约。例如，生成模型可能预测出照片级真实但物理上不可行的帧，从而使低层策略产生混淆。低层策略也可能对生成目标图像中细微的视觉伪影十分敏感。本文针对泛化问题的这两个方面展开研究，提供了一个接口，以有效地将语言条件的图像或视频预测模型与低层目标条件策略“粘合”在一起。我们的方法名为生成式分层模仿学习-Glue（Generative Hierarchical Imitation Learning-Glue，GHIL-Glue），它会过滤掉无法带来任务进展的子目标，并提升目标条件策略对含有有害视觉伪影的生成子目标的鲁棒性。在仿真和真实环境中的大量实验表明，GHIL-Glue 在若干利用生成子目标的分层模型上实现了 25% 的提升；对于仅使用单个 RGB 相机观测的策略，它在 CALVIN 仿真基准上达到了新的最先进水平。在物理实验中测试零样本泛化的 4 项语言条件操控任务里，GHIL-Glue 还在其中 3 项上优于其他通用机器人策略。

英文摘要：

Image and video generative models that are pre-trained on Internet-scale data can greatly increase the generalization capacity of robot learning systems. These models can function as high-level planners, generating intermediate subgoals for low-level goal-conditioned policies to reach. However, the performance of these systems can be greatly bottlenecked by the interface between generative models and low-level controllers. For example, generative models may predict photorealistic yet physically infeasible frames that confuse low-level policies. Low-level policies may also be sensitive to subtle visual artifacts in generated goal images. This paper addresses these two facets of generalization, providing an interface to effectively "glue together" language-conditioned image or video prediction models with low-level goal-conditioned policies. Our method, Generative Hierarchical Imitation Learning-Glue (GHIL-Glue), filters out subgoals that do not lead to task progress and improves the robustness of goal-conditioned policies to generated subgoals with harmful visual artifacts. We find in extensive experiments in both simulated and real environments that GHIL-Glue achieves a 25% improvement across several hierarchical models that leverage generative subgoals, achieving a new state-of-the-art on the CALVIN simulation benchmark for policies using observations from a single RGB camera. GHIL-Glue also outperforms other generalist robot policies across 3/4 language-conditioned manipulation tasks testing zero-shot generalization in physical experiments.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。