日常交互操作数据集

日常交互操作数据集由南佛罗里达大学计算机科学与工程系创建，专注于日常任务中的交互操作，如位置、方向、力和扭矩的记录。数据集包含32种日常动作的1593次试验和单独倾倒动作的1596次试验，旨在通过人类演示教会机器在变化环境中执行日常交互操作。该数据集支持任务导向的交互操作研究，适用于学习从演示中获取动作或力，以及动作识别和理解，对抓取研究也有潜在益处。

This daily interaction manipulation dataset was created by the Department of Computer Science and Engineering at the University of South Florida. It focuses on manipulation operations in daily tasks, involving the collection of data such as position, orientation, force and torque. The dataset includes 1593 trials covering 32 daily actions and 1596 trials solely dedicated to the pouring action. It aims to teach machines to perform daily interaction manipulation tasks in dynamic environments via human demonstrations. This dataset supports task-oriented interaction manipulation research, and is applicable to learning actions or forces from demonstrations, as well as action recognition and understanding, with potential benefits for grasping-related research.