DynaMo：面向视觉-运动控制的域内动力学预训练
DynaMo: In-Domain Dynamics Pretraining for Visuo-Motor Control
AI总结：
针对模仿学习处理高维视觉观测时数据效率低的问题，提出域内自监督方法DynaMo，通过联合学习潜在逆动力学与前向动力学模型学习视觉表示，在多环境多策略类别下显著提升下游模仿学习性能。
摘要：
模仿学习已被证明是训练复杂视觉运动策略的有力工具。然而，现有方法通常需要数百到数千个专家演示才能处理高维视觉观测数据。这种数据效率低下的一个关键原因是，视觉表示主要要么在域外数据上预训练，要么直接通过行为克隆目标训练。在这项工作中，我们提出了DynaMo，一种新的域内自监督视觉表示学习方法。给定一组专家演示，我们在图像嵌入序列上联合学习潜在逆动力学模型和前向动力学模型，在潜在空间中预测下一帧，无需数据增强、对比采样，也无需访问真实动作。重要的是，DynaMo不需要任何域外数据，例如互联网数据集或跨具身数据集。在六个模拟和真实环境组成的测试套件中，我们表明，与以往的自监督学习目标及预训练表示相比，使用DynaMo学习的表示显著提升了下游模仿学习性能。使用DynaMo带来的增益在Behavior Transformer、Diffusion Policy、MLP和最近邻等多种策略类别中均成立。最后，我们对DynaMo的关键组件进行了消融实验，并衡量了其对下游策略性能的影响。机器人视频可在https://dynamo-ssl.github.io 查看。
英文摘要：
Imitation learning has proven to be a powerful tool for training complex visuomotor policies. However, current methods often require hundreds to thousands of expert demonstrations to handle high-dimensional visual observations. A key reason for this poor data efficiency is that visual representations are predominantly either pretrained on out-of-domain data or trained directly through a behavior cloning objective. In this work, we present DynaMo, a new in-domain, self-supervised method for learning visual representations. Given a set of expert demonstrations, we jointly learn a latent inverse dynamics model and a forward dynamics model over a sequence of image embeddings, predicting the next frame in latent space, without augmentations, contrastive sampling, or access to ground truth actions. Importantly, DynaMo does not require any out-of-domain data such as Internet datasets or cross-embodied datasets. On a suite of six simulated and real environments, we show that representations learned with DynaMo significantly improve downstream imitation learning performance over prior self-supervised learning objectives, and pretrained representations. Gains from using DynaMo hold across policy classes such as Behavior Transformer, Diffusion Policy, MLP, and nearest neighbors. Finally, we ablate over key components of DynaMo and measure its impact on downstream policy performance. Robot videos are best viewed at https://dynamo-ssl.github.io
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。
