机器人多维力传感器标定Kalman滤波  

Kalman Filter for the Multi-Component Force/Moment Sensor of Robot Calibration

作　　者：许德章[1,2,3] 葛运建[1,2] 高理富[1] 

机构地区：[1]中国科学院合肥智能机械研究所,安徽合肥230031 [2]中国科学技术大学 [3]安徽工程科技学院

出　　处：《电子测量与仪器学报》2006年第1期92-97,共6页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金项目"数字运动员仿真"(编号:60343006)安徽省自然科学基金"基于力学信息和数字运动员人体仿真的运动指导系统研究"(编号:03042304)资助

摘　　要：在多维力传感器标定过程中,往往出现比较大测量噪声,零漂幅度较大,严重地限制了多维力传感器标定精度。鉴于Kalman滤波器在滤除系统随机干扰噪声方面良好效果,并考虑到在力传感器标定加载前,噪声信号便于测量的特点。本文从单维力传感器入手,把标定测量模型简化为一阶惯性和零阶保持器串联,并把传感器的输入/输出分为有载荷作用和无载荷作用两个状态,分别推导出输入/输出关系,获得单维力传感器状态和测量方程,并进一步推导出单维力传感器Kalman滤波算法。在合理假设基础上,再使单维力传感器标定Kalman滤波算法推广到多维力传感器。标定实验表明,在多维力传感器标定中,Kalman滤波有效地滤除了测量噪声,提高了标定精度。During calibrating for multi-component force/moment sensors, there are often measure noises that exhibit more draft range. Because it is difficult to cancel the drifts by analog circuit filters, the drifts badly restrict calibration precision of multi-component force/moment sensor. According to Kalman filtering fine effect of canceling drift, and convenience for measuring easily noise signals of the multi-component force/moment sensor, the paper simplified the single-component force sensor into a system for a series connection with a class inertial device and zero-holder. After the input and output of the single-component force sensor is divided into two states with load and without load, then the input and output relation of the single-component force sensor is derived respectively, finally the state and measure equations are inferred. Based on reasonable hypothesis, Kalman filtering for single-component force sensor is applied to the multi-component sensor. The Calibration experiment showed that filtering algorithm is successful in canceling drift, and improving effectively the calibrating precision of multi-component force/moment sensor.

关 键 词：多维力传感器 机器人 标定 KALMAN 滤波 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TN713[自动化与计算机技术—控制科学与工程]