Imitation Learning Demos

该数据集包含了50个针对“捡箱子”和“插笔”任务的演示，以及80个针对“收集玩具”任务的演示。在数据集中，机器人的状态由18个维度表示，而远程操作者则通过逆运动学提供了18维的指令。具体到规模上，“捡箱子”和“插笔”任务各有50个演示，“收集玩具”任务则有80个演示。这些数据集的任务旨在进行模仿学习。

This dataset includes 50 demonstrations for each of the "box picking" and "pen inserting" tasks, plus 80 demonstrations for the "toy collection" task. Specifically, both the "box picking" and "pen inserting" tasks have 50 demonstrations each, while the "toy collection" task has 80 demonstrations. In this dataset, the robot's state is represented as an 18-dimensional vector, and the remote operator provides 18-dimensional control instructions via inverse kinematics. All these task demonstrations are intended for imitation learning.