Penentuan Orientasi dan Translasi Gerakan UAV menggunakan Data Fusion berbasis Kalman Filter

Abstract
In an unmanned aircraft vehicle, a navigation system is needed to calculate its orientation and translation. The navigation system can utilize data from the accelerometer, gyroscope, magnetometer, and GPS. The orientation can be precisely calculated from the accelerometer and magnetometer data when the sensor is in a static state. Meanwhile, under dynamic conditions, the orientation can be more precisely calculated from the gyroscope data. In order to obtain the robust navigation system, a data fusion based on Kalman filter is built to calculate the orientation from the accelerometer, gyroscope, and magnetometer. The Kalman filter trusts more in the data from the accelerometer and magnetometer when the UAV is static and trusts more in to the gyroscope data when the UAV is in dynamic conditions. Meanwhile, the UAV translation is obtained by performing data fusion of the accelerometer data with location data from the GPS sensor. The Kalman filter combines data from the accelerometer and GPS when available, otherwise trusts in data from the accelerometer only. The trust level shifting is done by changing the measurement noise covariance. The data fusion based on Kalman filter provides more accurately the orientation and translation data. The orientation as a result of the calculation from the gyroscope has an average error of 18.12%, while the orientation as a result of the accelerometer and magnetometer has an error of 1.3%. By using Kalman filter-based data fusion, the error of the orientation decreases to 0.87%