Research On Integrated Yaw Estimation Method Using Inertial And Magnetic Sensors

The yaw,pitch,and roll angle compose a set of Euler angle,which describes the orientation of an object completely.Orientation estimation is essential for many applications including navigation,motion tracking,virtual reality,assisted driving,and so on.MEMS gyroscopes,MEMS accelerometers,and AMR magnetometers have become commonly used sensors in orientation estimation because of small size,low power consumption,and low cost.In this thesis,the integrated yaw estimation method using inertial and magnetic sensors is studied and the main work involves compensation of gyroscope random error,calibration of triaxial magnetometer and integrated yaw estimation algorithm.The main innovation points of this thesis are as follows.1)A new risk sensitive unscented Kalman filter algorithm suitable for nonlinear model is implemented.The risk sensitive unscented Kalman filter has strong robustness,and the nonlinear model can realize online estimation of error parameters.To combine the filter and the nonlinear model,the augmented state vector method is applied to deal with the non-additive noise term.To improve the numerical stability,QR decomposition and Cholesky factor updating are applied to realize a square root form.In addition,the scalar risk parameter is extended to matrix form to improve the flexibility.In experiments,different error models and filters are used to deal with the gyro random errors.And the results show that the new risk sensitive unscented Kalman filter combined with the nonlinear model achieves the best compensation effect.2)An equivalent two step algorithm based on geometric meaning is proposed for the calibration of triaxis magnetometer.Calibration of magnetometer is necessary to compensate deterministic errors in geomagnetic field measurement.The proposed algorithm divides the calibration into equivalent sensor frame calibration and equivalent misalignment calibration.It uses the ellipsoid method and the principal component analysis respectively to perform the equivalent sensor frame calibration and equivalent misalignment calibration.And it corrects the sign inversion and nonorthogonality defects in misalignment calibration.Experiments results show that the proposed algorithm can work independently and obtain complete error parameters.3)A new measurement equation,namely the stable yaw measurement equation is proposed,and the way of measurement update is improved.In yaw estimation algorithms based on multi-sensor information fusion,the conventional strategy to handle magnetic disturbances is reducing the magnetometer measurement weight,which results in ineffective correction for the yaw estimation.To address this problem,the stable yaw measurement equation is constructed based on carrier status.Two forms of the equation are derived,and the first form is used to improve the yaw estimation algorithm.To describe the vehicle status,the correlations between outputs from different sensors are analyzed,and a vehicle status marker is constructed.In conventional algorithms,magnetic disturbances can impact attitude estimation through the magnetometer measurement update.To address this problem,the improved way of measurement update performs the magnetometer measurement update separately,and restores attitude estimation according to the invariable attitude condition.Experiments results show that the stable yaw measurement equation is an effective supplement to the magnetometer measurement equation in magnetically disturbed environments,and the improved measurement update way makes attitude estimation immune to inaccurate geomagnetic field measurements.Both of them effectively enhance the ability of the yaw estimation algorithm to handle magnetic disturbances...