Aiming at the problem of non-Gaussian noise attitude estimation in multi-sensor fusion attitude calculation

a Kalman filter based on entropy criterion is proposed. The minimum entropy criterion filtering method can reduce the estimation error better. The mathematical model established by the quaternion derivation of the accelerometer can estimate the direction of the gyro well

which not only reduces the calculation amount of the CPU

but also better corrects the data of the gyro. In order to verify the feasibility of the algorithm

the IMU module is used to compare the steepest descent algorithm and the classical Kalman filtering algorithm. The experimental results show that the Kalman filtering algorithm based on entropy criterion can better estimate the attitude direction and remove noise indoors. It has high precision and reliability.