Multi-sensor fusion based unscented attitude estimation method for MAVs

摘要

MAVs (Micro Aerial Vehicles) usually adopts the low cost of MEMS (Micro-electromechanical Systems) devices for attitude calculation, which has the feature of low precision and high noise, and error accumulated over time quickly. Aiming at this problem, the attitude estimation method based on UKF (Unscented Kalman Filter) is designed. The algorithm treats attitude as filter state, and update state with measured angular rate. Then update attitude state by gravitational and geomagnetic data for inhibiting the error divergence. The simulation results show that the attitude estimation accuracy is up to 0.027 radians around which is better than EKF, and the calculation speed of single calculation is approximately 0.0005s, and it is far less than sampling interval which equals to 0.1s, meet the real-time requirements of operation.