基于EM算法的工业机器人抓取控制研究  

Research on grasping control of industrial robot based on EM algorithm

作　　者：卢晓玲 孙自昌 Lu Xiaoling;Sun Zichang(Mechanical and Electrical Engineering,Anhui Automobile Vocational and Technical College,Hefei 230601,China)

机构地区：[1]安徽汽车职业技术学院机电工程系,安徽合肥230601

出　　处：《湖南文理学院学报（自然科学版）》2021年第3期20-23,75,共5页Journal of Hunan University of Arts and Science(Science and Technology)

基　　金：安徽省高校科学研究重点项目(KJ2019A1309)。

摘　　要：抓取是机器人的基本操作任务,通过待抓取物体的位姿变化对机器人关节数据自适应调整,提高机器人物体抓取成功率。本文基于EM算法的混合正态分布模型建立待抓取物体位姿观测变量和机器人关节变量之间的映射关系,并采用UR3机器人进行抓取试验。结果表明,采用该方法进行抓取控制只是在样本训练集的边缘抓取失败,抓取成功率为95.5%。基于EM算法的混合正态分布模型可以有效地对工业机器人抓取进行控制,具有比较高的抓取成功率。Grasping is the basic operation task of a robot.The robot joint data can be adjusted adaptively by the change of the position and posture of the object to be grasped,so as to improve the success rate of the robot object grasping.Based on the mixed normal distribution model of EM algorithm,the mapping relationship between the pose observation variables of the object to be grasped and the joint variables of the robot are established,and the UR3 robot is used for grasping experiments.The results show that only the edge of the sample training set fails to grasp with this method,and the success rate of grasping is 95.5%.The hybrid normal distribution model based on EM algorithm can effectively control the grasping of industrial robot,and has a high success rate of grasping.

关 键 词：EM算法 工业机器人 混合正态分布 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]