灵巧臂-手系统柔顺操作数据

本数据集于2022年11月至2025年10月在重庆大学完成，基于 UR5 机械臂 + 灵巧手，并同时安装安装 Hypersen六维力传感器，在真实多场景、多任务条件下实地采集。基于灵巧臂-手系统对抓、捏、握、拧、旋、推、拉、按、插、拔等10种基础操作模式开展实验和对针对杯子、衣物、线缆、接头、螺栓、门把手、阀门、工具（锤子、电动螺丝刀）等10种典型物体开展抓、捏、握、推、拉、按、插、拔等操作实验，同步采集关节角度、灵巧手力觉数据、多模态触觉数据以及腕部力觉数据。本数据集可用于动作基元理解、抓取分析、力觉融合、多模态学习以及机器人操作任务的研究和模型训练。This dataset was collected at Chongqing University from November 2022 to October 2025. The data acquisition platform comprises a UR5 robotic arm paired with a dexterous hand, and a Hypersen 6-axis force/torque sensor installed simultaneously. All data was collected in situ under realistic multi-scenario and multi-task conditions. Using the dexterous arm-hand system, experiments were conducted on 10 basic manipulation modes including grasping, pinching, holding, twisting, rotating, pushing, pulling, pressing, inserting and extracting. Additional experiments were carried out on 10 typical objects including cups, clothing, cables, connectors, bolts, door handles, valves, and tools (hammer, electric screwdriver), covering the aforementioned manipulation operations. During the experiments, joint angles, dexterous hand force data, multimodal tactile data and wrist force data were synchronously collected. This dataset can be applied to research and model training for tasks such as action primitive understanding, grasp analysis, force fusion, multimodal learning and robotic manipulation research.

搜集汇总

背景与挑战

背景概述

该数据集于2022年11月至2025年10月在重庆大学采集，基于UR5机械臂和灵巧手系统，针对10种操作模式和10种典型物体，同步记录了关节角度、力觉及多模态触觉数据。它可用于机器人操作任务中的动作基元理解、抓取分析、力觉融合和多模态学习研究与模型训练。