Multi-Rotor UAV Attitude Calculation Based on Extended Kalman Filter

This paper designs a new observer of the attitude fusion algorithm which is applicable to small unmanned aerial vehicles (UAVs) using MEMS sensors in non-stationary environment. When the UAV is under accelerative environment, the accelerometer degrades the accuracy of estimated attitude. Attitude heading reference system (AHRS) propagates the attitude by integrating gyroscope output and determines the attitude with gravity and magnetic field measurements from the accelerometer and magnetometer. The accelerometer output reflects not only gravity but also the acceleration of the vehicle. In this case, the conventional accelerometer-based method gives inaccurate attitude information and degrades system performance. Therefore, a new decoupling observer is designed, which is dependent on the relationship between UAV attitude angle and motion acceleration. The simulation results show that the proposed method can improve the attitude accuracy without increasing the computational complexitywhich verifies the effectiveness of the method. Meanwhile the maximum error drops by 2 orders of magnitude.