Pose Estimation of a Drone Using Dynamic Extended Kalman Filter Based on a Fuzzy System

When conducting pose estimation of a drone in an external environment, inertial navigation systems (INSs) and global navigation satellite systems (GNSSs) are often used. Filter-based extended Kalman filter (EKF), unscented Kalman filter (UKF) and particle filter (PF) based on sampling method are used for estimating pose more accuracy. This study attempted to improve the accuracy of EKF using a fuzzy system, which controls covariance according to environment. In order to test the algorithm, ROS, Gazebo, and Pixhawk offboard firmware were used.