基于机械臂灵巧手智能数据采集系统的设计与分析  

Design and Analysis of Intelligent Data Acquisition System Based on Robotic Hand

作　　者：杨青丰 冯宝林[1,2] 李露[1,2] 施云高 孙鹏[1,2] 毛吴俊 YANG Qingfeng;FENG Baolin;LI Lu;SHI Yungao;SUN Peng;MAO Wujun(Hefei Institute of Material Science,Chinese Academy of Sciences,Hefei 230031,China;Institute of Advanced Manufacturing Technology,Hefei Institute of Material Science,Chinese Academy of Sciences,Changzhou 213164,China)

机构地区：[1]中国科学院合肥物质科学研究院,安徽合肥230031 [2]中国科学院合肥物质科学研究院先进制造技术研究所,江苏常州213164

出　　处：《机械与电子》2019年第7期75-80,共6页Machinery & Electronics

摘　　要：在非结构环境中机器人对不同形状、重量物体的有效抓取效率低。针对这一现状,自主设计了一套基于机械臂灵巧手的智能数据采集系统平台,该平台由Kinect2.0摄像机、BH8-282三指灵巧手、UR5六自由度机械臂等设备组成,通过对目标物体的自动识别和定位,自主运动规划路线完成对目标物体的抓取动作,并得到抓取目标物的视觉和触觉信息。实验表明,该平台可以在无人监督的情况下,完成对目标物体的有效抓取,并实现对视觉和触觉数据的完全自动化采集。实验过程中对3589组抓取目标物数据分析,未抓住目标物的比例为15.41%,抓稳的比例为42.91%,未抓稳的比例为41.86%,总体实验效果较好。In unstructured environment, robots can not effectively capture different shape and weight objects efficiently. In view of this situation, a platform of intelligent data acquisition was designed based on the Dexterous Manipulator hand, which consists of a Kinect 2.0 camera, BH8-282 three-finger dexterous hand, UR5 six-DOF robot arm and so on. According to the automatic identification and position of the target object, the platform can complete the crawl action and get the target object's visual and tactile information through the autonomous movement planning route. Experiments show that the platform can effectively capture the object under the condition of unsupervised and complete automatic collection of visual data and tactile data. By analyzing the experimental data of 3 589 groups, it was found that the proportion of grasping failure was 15.41%, the ratio of stable grasping was 42.91%, the proportion of unstable grasping was 41.86%. In short, the overall experimental results was good.

关 键 词：机器人 机械臂 触觉 自动化采集 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]