PoseIt

PoseIt是由卡内基梅隆大学创建的一个多模态数据集，专注于研究抓取稳定性分析中的持有姿势。该数据集包含1840个抓取数据点，来自26个不同对象，每个对象有16种不同的持有姿势。数据收集过程包括抓取物体、移动到持有姿势以及摇动物体以检查其稳定性，同时收集RGB-D、触觉、力、扭矩、关节角度、关节速度和夹持器力等信息。PoseIt旨在通过视觉和触觉数据，帮助研究不同持有姿势对抓取稳定性的影响，并应用于机器人抓取任务中，解决抓取稳定性的预测问题。

PoseIt is a multimodal dataset developed by Carnegie Mellon University, focusing on the investigation of holding poses in the context of grasp stability analysis. This dataset consists of 1,840 grasp data points sourced from 26 distinct objects, with 16 different holding poses for each individual object. The data collection workflow includes grasping the target objects, moving them to the specified holding poses, and shaking the objects to verify their stability, while concurrently collecting multiple data modalities such as RGB-D, tactile signals, force readings, torque measurements, joint angles, joint velocities, and gripper force. PoseIt is designed to facilitate research on the influence of diverse holding poses on grasp stability using visual and tactile data, and to be applied in robotic grasping tasks to address the problem of grasp stability prediction.