Attitude Estimation Method for Smart Watch Using External Disturbance Model

In this paper, we propose an attitude estimation method using a disturbance model. The proposed method has two parts. One is estimation external disturbance. The magnetic disturbance is augmented as states and estimated by filters. Disturbances are estimated by comparing sensor readings to criteria such as gravity vectors. The proposed method subtracts the estimated disturbance from the sensor measurement and uses it for filter update to maximize the use of measurement information. The second part is that The magnetometer measurement update is performed only on the desired state variable. In relation to magnetometer measurements and attitude, measurements affect not only yaw but also roll and pitch. The proposed method uses a partial-update Schmidt Kalman filter to minimize the effect of magnetometer measurements on roll and pitch. To evaluate the performance of the proposed method, sensor data collected during outdoor exercise by a tester wearing a smart watch was used. Experimental results indicate that the proposed method improves estimation performance.