Application of quaternion-based extended Kalman filter for MAV attitude estimation using MEMS sensors

In this paper, a new quaternion-based extended Kalman filter algorithm was proposed to improve the accuracy of attitude estimation of micro aerial vehicles based on the MEMS sensors such as gyroscope, accelerometer and magnetometer. Attitude quaternion errors and drift bias of gyroscope were selected to construct a state vector, and the state equation was established based on attitude quaternion error differential equation and stochastic error model of gyroscope. The modified Gauss-Newton algorithm was used to convert the outputs of sensors to quaternion by which the measurements of Kalman filter were obtained through making combination of attitude quaternion from gyroscope signals, which significantly reduces the influences of maneuvering acceleration on attitude estimation. Experimental results show that the maximum estimated errors were less than 0.22% in static and the estimated quaternion could well track its true values dynamically. The algorithm is proved to be effective at improving the accuracy of attitude estimation.