基于熵准则的卡尔曼滤波算法的姿态估计

Attitude estimation of Kalman filtering algorithmbased on entropy criterion

针对多传感器融合姿态解算中非高斯噪声姿态方向估计的问题,提出一种基于熵准则的卡尔曼滤波器的设计,最小熵准则的滤波方法能更好地减少估计误差,利用基于加速度计四元数推导建立的数学模型,能够很好地估计解算陀螺的方向,这样既减少了CPU的计算量,又能更好地校正陀螺的数据。为了验证算法的可行性,利用惯性测量单元(IMU)模块对比了最速下降算法、经典的卡尔曼滤波算法,实验结果表明:基于熵准则的卡尔曼滤波算法在室内能更好地估计姿态方向和去除噪声,且具有较高的精度和可靠性。

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.