Dexterous Manipulation Tasks

该数据集涵盖了多指机械手执行的各种灵巧操作任务，特别关注通过无模型的深度强化学习方法来学习行为。对于每个任务，通过动态教学收集了演示数据，数据集包含了Dclaw和Allegro两种机械手的结果。该数据集在多种硬件设置下评估了多项任务，旨在研究使用深度强化学习来学习灵巧的操作行为。

This dataset covers a wide range of dexterous manipulation tasks executed by multi-finger robotic hands, with a special focus on learning behaviors through model-free deep reinforcement learning approaches. For each task, demonstration data was collected via dynamic teaching. The dataset includes results from two types of robotic hands: Dclaw and Allegro. This dataset evaluates multiple tasks across diverse hardware setups, with the goal of investigating the use of deep reinforcement learning for learning dexterous manipulation behaviors.