Research On Attitude Estimation Using Low-cost Inertial Measurement Unit

Attitude estimation plays important roles in military,aerospace,UAV control,virtual reality,augmented reality,robot control,human attitude measurement,and so on.In recent years,with the development of Micro Electro Mechanical System(MEMS),the MEMS inertial measurement units(IMU,including three-axis gyroscope,accelerometer and magnetometer)with higher accuracy are available.The MEMS IMU has many advantages,such as low price,small volume and light weight.Therefore,it has been widely used in attitude estimation.When the MEMS gyroscope outputs are used for attitude estimation,the attitude drift will occur due to large measurement noises and gyroscope bias instability.To tackle this problem,accelerometer measurements can be used to correct horizontal attitude(roll and pitch).However,it cannot be directly used due to the influence of interference acceleration caused by external forces.In addition,the magnetometer measurements can also be used to correct the heading angle.Nevertheless,it is prone to be disturbed by the magnetic field of the carrier itself and the environment.Thus,it is necessary to calibrate it before usage.This paper studied the key techniques of attitude estimation using MEMS inertial measurement unit.The main contents are listed as follows.At first,the attitude estimation algorithm based on the error-state Kalman filtering is studied.This algorithm fuses the gyroscope,accelerometer and magnetometer measurements to estimate the attitude.Then,the validity of the algorithm is verified by numerical simulations.The magnetometer calibration is realized based on the maximum likelihood estimation method.Inspired by the first-order autoregressive model and the perspective of frequency domain matching,a prediction and compensation method of interference acceleration is proposed based on the second-order AR model.The proposed method is combined with the threshold method to improve the attitude estimation accuracy.Finally,the validity of the proposed method is verified using real data collected from Xsens MTi-G700.