基于惯性数据融合的串联机械臂运动状态估计方法

Motion state estimation method for serial manipulator based on inertial data fusion

在串联机械臂运动跟踪中,将惯性测量单元(IMU)的运动信息映射到机械臂运动链上,针对IMU测量时的安装误差和随机噪声会影响到测量信息准确性的问题,提出了一种基于IMU数据融合的运动状态估计方法。该方法首先对传感器与旋转关节坐标系进行校准,再通过融合IMU中陀螺仪和加速计信息,建立了基于扩展卡尔曼滤波(EKF)算法的运动姿态最优估计模型。与机械臂实际输出的角度信息相比,所提方法的平均角度和位姿均方根误差(RMSE)为0.19°和0.45 mm,均优于单目相机惯性传感器联合标定方法。实验结果表明:该方法能够精确地估计机械臂运动状态,并且具有精度更高且运动空间不受限制的优点。

In serial manipulator motion tracking,motion information of the inertial measurement unit(IMU)is mapped to the manipulator motion chain.Aiming at the problem that the installation error and random noise during IMU measurement will affect the accuracy of the measurement information,a motion state estimation method based on IMU data fusion is proposed.The method calibrates the coordinate system of the sensor with the rotating joint firstly.Then,an optimal motion state estimation model is built based on extended Kalman filtering(EKF)algorithm by fusing the information from gyroscope and accelerometer of IMU.Compared with the actual output angle information of the manipulator,the proposed method has an average angle and root mean square error(RMSE) of pose 0.19°and 0.45 mm,respectively,which are both prior to those of monocular camera-inertial sensor calibration method.The experimental results show that the method can estimate the manipulator motion state accurately,and has the advantages of higher precision and unrestricted motion space.