Adaptive Filter Design with Unconstrained Optimization For Heading Angle Oscillation Problems

In navigation systems, sensors that measure acceleration and magnetic field are mainly used to obtain direction angle information. Sensors produce noisy output data that changes over time due to disruptive factors (vibration, shock, magnetic field change) of the system or environment in which they are located. In order to obtain high-accuracy sensor data, it is necessary to develop a good digital filter as well as a good sensor calibration. In this study, an adaptive filter design was carried out to eliminate corruptive effects. Punishment/reward coefficients in the filter are found using the Steepest Descent Optimization method, which is one of the Unconstrained Optimization methods.