We propose a method for improving the prediction accuracy of learned robot dynamics models on out-of-distribution (OOD) states. We achieve this by leveraging two key sources of structure often present in robot dynamics: 1) sparsity, i.e., some components of the state may not affect the dynamics, and 2) physical limits on the set of possible motions, in the form of nonholonomic constraints. Crucially, we do not assume this structure is known \textit{a priori}, and instead learn it from data. We use contrastive learning to obtain a distance pseudometric that uncovers the sparsity pattern in the dynamics, and use it to reduce the input space when learning the dynamics. We then learn the unknown constraint manifold by approximating the normal space of possible motions from the data, which we use to train a Gaussian process (GP) representation of the constraint manifold. We evaluate our approach on a physical differential-drive robot and a simulated quadrotor, showing improved prediction accuracy on OOD data relative to baselines.

翻译：我们提出了一种方法，用于提升学习型机器人动力学模型在分布外（OOD）状态下的预测精度。该方法利用机器人动力学中常见的两种关键结构信息：1）稀疏性，即状态中的某些分量可能不影响动力学；2）物理限制，即非完整约束对可能运动集合的约束。关键在于，我们并不假定这些结构先验已知，而是从数据中学习它们。我们采用对比学习获取一种距离伪度量，用以揭示动力学中的稀疏模式，并将其用于缩减动力学学习中的输入空间。随后，通过从数据中近似可能运动的法空间，我们学习未知的约束流形，并基于此训练约束流形的高斯过程（GP）表示。我们在物理差分驱动轮式机器人和仿真四旋翼飞行器上评估了该方法，结果表明相较于基线方法，模型在分布外数据上的预测精度得到了提升。