Adaptive Manipulation Demonstrations

该数据集包含了9类可变形物体的自适应操作演示，每类物体有20个演示样本。这些数据旨在用于训练自适应操作策略。此外，该数据集不仅包括模拟演示，还包括真实世界的演示，这些演示被评估为在物体状态变化下训练自适应策略的有效性。规模上，数据集涵盖了9类物体，每类物体有20个演示，以及在实际试验中每类物体35个专家演示。该数据集的任务是学习自适应操作策略。

This dataset contains adaptive manipulation demonstrations for 9 categories of deformable objects, with 20 demonstration samples per category. These data are designed for training adaptive manipulation policies. Furthermore, the dataset encompasses both simulated and real-world demonstrations, which have been validated for their efficacy in training adaptive policies under varying object states. In terms of scale, apart from the 20 demonstrations per category previously mentioned, the dataset also includes 35 expert demonstrations per category in real-world experiments. The core task of this dataset is to learn adaptive manipulation strategies.