State estimation for UAVs using sensor fusion

In this paper an improved approach is presented for the state estimation of unmanned aerial vehicles (UAVs). The three-loop technique is based on Extended Kalman Filters. From them EKF1 solve the quaternion based orientation (attitude) estimation using IMU and magnetometer. EKF2 improves the attitude estimation if GPS information is present. The third filter EKF3 determines the remaining state variables including the biases in an external loop if GPS measurement is available and tolerates the large difference between IMU and GPS frequencies. The method can be applied for state estimation of any type of vehicles. It is especially useful for vehicles having large (2-3 G) acceleration changes typical for UAVs. The results can later be used for identification of the nonlinear dynamic model of vehicles (control surface, thrust and inertia effects) building the basis for advanced control.