An adaptive attitude algorithm based on a current statistical model for maneuvering acceleration

A current statistical model for maneuvering acceleration using an adaptive extended Kalman filter (CS-MAEKF) algorithm is proposed to solve problems existing in conventional extended Kalman filters such as large estimation error and divergent tendencies in the presence of continuous maneuvering acceleration. A membership function is introduced in this algorithm to adaptively modify the upper and lower limits of loitering vehicles' maneuvering acceleration and for real-time adjustment of maneuvering acceleration variance. This allows the algorithm to have superior static and dynamic performance for loitering vehicles undergoing different maneuvers. Digital simulations and dynamic flight testing show that the yaw angle accuracy of the algorithm is 30% better than conventional algorithms, and pitch and roll angle calculation precision is improved by 60%. The mean square deviation of heading and attitude angle error during dynamic flight is less than 3.05°. Experimental results show that CS-MAEKF meets the application requirements of miniature loitering vehicles.