Comparison of EKF, EKF2 and UKF in a loosely coupled INS/GPS integration

---

Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye

Tezin Onay Tarihi: 2018

Öğrenci: HANDE JANE

Danışman: BUYURMAN BAYKAL

Özet:

Developing effective navigation solutions in land, sea, air, and space travels is an ever increasing necessity maintaining its validity. Nowadays, although a navigation aid called satellite based Global Positioning System (GPS) is used to develop a navigation solution, standalone GPS cannot achieve the desired accuracy in some applications. For this reason, GPS measurements are integrated with the Inertial Navigation System (INS) to develop an enhanced accurate integrated navigation solution. Extended Kalman Filter (EKF) is used widely as an estimation algorithm in the INS/GPS integrated systems. In this thesis, instead of EKF, unscented Kalman Filter (UKF) and Second Order Extended Kalman Filter (EKF2) shall be used and comparison is made between the results in terms of the accuracy of the navigation systems, sensitivity to Inertial Measurement Unit (IMU) noise, effect of changing process noise covariance matrix and complexity for GPS available situation. According to experimental results, it is observed that UKF increased the accuracy of the navigation system with respect to EKF and EKF2, as mentioned in the literature, EKF2 has better accuracy performance than EKF. In addition to this, EKF is more sensitive to additive noise on IMU measurements with respect to UKF. Also, increasing the noise on IMU measurements forces the EKF2 to diverge quickly. UKF can work with wide range of noise levels with respect to the other algorithms. EKF, UKF and EKF2 complexity is rising, respectively. Also, accuracy of estimation algorithms are compared for GPS interference (outage or outlier). UKF has the best performance with respect to EKF and EKF2 when GPS signal is lost and although EKF2 introduces an increase in the performance according to EKF without GPS interruption, it is not robust to GPS outages and/or GPS outlier. For nonlinear problems, EKF is a good solution for GPS outlier situation as a result of poor robustness of UKF.