基于共轭梯度法和互补滤波相结合的姿态解算算法

Attitude Estimation Based on Conjugate Gradient and Complementary Filter

为了提高姿态解算精度,提出了一种基于共轭梯度和互补滤波相结合的多传感器数据融合策略。系统采用四元数方法进行姿态解算。利用加速度计和磁强计的输出数据,通过共轭梯度方法对姿态四元数进行寻优估计,再将其和利用陀螺仪输出数据更新的四元数进行互补滤波,解算出姿态角。实验测试表明,这种融合策略使姿态检测系统静态性能和和动态性能均有所提高,尤其在姿态剧烈变化时,其性能明显优于卡尔曼滤波和梯度下降法。

In order to improve the precision of attitude estimation, a novel attitude estimation algorithm based on conjugate gradient and complementary filter is proposed. The quaternion representation used in the proposed system for attitude estimation, is optimized based on conjugate gradient algorithm with magnetometer and accelerometer data,and then fused with the updated quaternion from gyroscope data based on complementary filter. Finally, experiment results verify that the proposed algorithm based on conjugate gradient and complementary filter improves the static and dynamic properties of attitude estimation system, and has better performance than the Kalman filter and gradient descent filter especially when the attitude has drastic changes.