Unscented Kalman Filter (UKF) - Doctoral Dissertation on Unscented Transformation Kalman Filtering

In this doctoral dissertation, the author provides a comprehensive introduction to the concept of Unscented Kalman Filter (UKF) and its applications across various domains. By improving upon traditional Kalman filtering methods, the UKF demonstrates superior capability in handling nonlinear systems and non-Gaussian noise. The algorithm employs an unscented transformation technique that propagates carefully selected sigma points through the nonlinear system, preserving mean and covariance statistics more accurately than linearization approaches. This dissertation conducts in-depth research on the UKF's mathematical model and algorithmic structure, including detailed explanations of key functions such as sigma point selection, state prediction, measurement update, and covariance management. The research presents extensive experimental results using MATLAB/Python implementation examples that demonstrate the UKF's performance in trajectory estimation, sensor fusion, and navigation applications. Through detailed analysis of real-world case studies and comparative performance evaluations, readers gain profound understanding of UKF principles and practical implementation techniques, providing valuable reference for research and applications in related fields including autonomous systems, signal processing, and control engineering.