Magnetometer-Based Attitude Determination Extended Kalman Filter and Optimization Techniques

Magnetometer-based attitude determination algorithm has been studied for decades owing to its advantages. This article aims to develop and fine-tune the magnetometer-based attitude determination method by introducing two new observation models for the extended Kalman filter (EKF). The optimization techniques were purposed to increase accuracy and improve convergence. The Monte Carlo simulation, including disturbances and a moment of inertia estimation error, was held to evaluate the performance of each EKF by the innovation vector and convergence. The significant effect of singularity on the reliability of the estimator was discovered and resolved by optimization. The study concluded that the attitude representation observation EKF with optimizations reduced the convergence time and provided reliability.