An Adaptive Extended Kalman Filter for Attitude Estimation Using Low-Cost IMU from Motor Vibration Disturbance

In the challenging environment of global navigation satellite systems (GNSS), in order to improve the accuracy of navigation, it is important to improve the accuracy of attitude calculation, which is significantly affected by various sensor noises and attitude calculation methods. The noise of sensor output will become larger with the increase of motor speed. In this paper, four algorithms including complementary filter method (CF), gradient descent algorithm (GDA), and extended Kalman filter method (EKF) are introduced for attitude estimation when the motor speed changes. Besides, in order to resist the influence of motor vibration, an adaptive factor is introduced into the EKF for attitude estimation. The result shows that the average value of adaptive extended Kalman filter (AEKF) solution error is about 0.13° and the variance is about 0.115°, which is greatly reduced compared with the other three methods.