基于模糊卡尔曼滤波的姿态估计  

Pose Estimation Based on Fuzzy Kalman Filtering

作　　者：黄伟 田明[1] 冯进良[1] 王菲[1] HUANG Wei;TIAN Ming;FENG Jinliang;WANG Fei(School of Opto-Electronic Engineering,Changchun University of Science and Technology,Changchun 130022)

机构地区：[1]长春理工大学光电工程学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2023年第2期120-127,共8页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：科技部国家重点研发计划项目(KYC-JC-XM-2018-157)。

摘　　要：在人机控制交互领域,通常通过惯性测量单元(IMU)采集手臂姿态原始信息并进行解算融合,从而达到获取人体手臂运动姿态以达到控制机械臂的目的。基于陀螺仪的动态特性和加速度计、磁力计的静态性能,考虑引入卡尔曼滤波进行姿态融合解算,由于卡尔曼滤波模型中的量测过程噪声协方差矩阵的统计特性难以准确获取,提出一种基于T-S模糊控制的卡尔曼滤波姿态估计方法,实现量测协方差矩阵的实时调整。基于ZYNQ7020搭建实验平台对手臂姿态信息进行融合解算,融合解算后的姿态角平均误差和均方误差在0.06和0.16以内,相比卡尔曼滤波降低了92.52%、89.44%,有效提高了姿态角解算精度。In the field of human-computer control interaction,the original information of the arm posture is usually collected by the inertial measurement unit(IMU)and the solution and fusion are carried out to achieve the purpose of obtaining the human arm movement posture to achieve the purpose of controlling the robotic arm.Based on the dynamic characteristics of the gyroscope and the static performance of the accelerometer and magnetometer,considering the introduction of Kalman filter for attitude fusion solution,since the statistical characteristics of the noise covariance matrix of the measurement process in the Kalman filter model are difficult to obtain accurately,a Kalman filter attitude estimation method based on T-S fuzzy control is proposed to realize the real-time adjustment of the measurement covariance matrix.Based on ZYNQ7020,an experimental platform was built to fuse and solve the attitude information of the arm,and the average error of the attitude angle and mean squared error after the fusion solution were within 0.06 and 0.16,which were reduced by 92.52%and 89.44%compared with Kalman filtering,which effectively improved the attitude angle solution accuracy.

关 键 词：卡尔曼滤波 模糊控制 FPGA 人机交互 IMU 

分 类 号：TH89[机械工程—仪器科学与技术]