Integrity on Low-cost INS/GNSS/Odometer Tightly Coupled Integration using Extended and Unscented Kalman Filter

For navigation, perception, control, and all-around safe operation, autonomous vehicles ought to know its accurate location and orientation in any kind of weather and traffic. Because autonomous vehicles operate under a wide range of conditions, several sensors are utilized to assure the necessary performance. The Extended Kalman filter (EKF) and Unscented Kalman filter (UKF) are utilized as sensor fusion algorithms to fuse low-cost INS/GNSS/Odometer. Nevertheless, there are a number of sources of error exist. It must be limited by the acceptable amount of risk for the application. Therefore, in this paper, we focus on Horizontal Protection Level (HPL) and choose the kSigma algorithm to compute it. By using tightly coupled integration, the comparison between EKF and UKF in the sense of HPL is analyzed. First, we analyzed the difference RMS position error between EKF and UKF. Then, the result both algorithms in Horizontal Position Error (HPE) subject to HPL are investigated. Overall, the UKF solution has a better result in accuracy and integrity than EKF. It indicated with lower percentage of epoch UKF solution that satisfied HPE < HPL condition than EKF and lower RMS position error.