This paper presents the results of a quaternion based extend Kalman filter (EKF) and complementary filter for ArduPilotMega (APM) attitude estimation. In addition, a new method to get the measurement noise covariance matrix R is proposed. Experimental results show that the two algorithms can meet the requirements, but the complementary filter can yield better performance than EKF.

This article presents use the BP neural network to do the attitude angle Information fusion for the Attitude and Heading Reference System (AHRS). In addition, use the more mature BP neural network in a different Area. The algorithm has been applied to the attitude control of a small UAV.