Improving Orientation Estimation Accuracy Using Dynamic Data Fusion of Multiple Sensors

The determination of UAV orientation is a challenging task. This is usually performed by a data fusion from accelerometers, gyroscopes, and magnetometers using a Kalman filter. However, the data fusion of these three-sensor does not always provide accurate orientation data due to various UAV conditions, such as static and dynamic states. Therefore, it is necessary to develop a dynamic data fusion approach that takes these conditions into account. UAV state determination is also an area of ongoing development. Numerous studies have been carried out to precisely assess the sensor's condition. However, the approach involves a relatively intricate algorithm, posing challenges for implementation on microprocessor-based systems. In this study, the determination of dynamic or static conditions is performed by measuring the variance of acceleration data and the total linear acceleration data. The use of this dynamic data fusion method can improve the accuracy of orientation estimation, especially for roll and pitch angles.