提出了一种基于四元数扩展卡尔曼滤波(QEKF)数据融合算法,应用于四旋翼飞行器的姿态估计。首先,用四元数表示航向姿态参考系统(AHRS)中的坐标旋转关系,获取飞行器的姿态信息;然后,结合扩展卡尔曼滤波算法,融合低成本的MPU9250中陀螺仪、加速度计、磁力计传感器的9轴原始输出数据,对飞行器的真实姿态进行跟踪估计;最后,通过MATLAB实验仿真,与互补滤波(CF)和梯度下降法(GD)算法进行比较,实验证明QEKF对姿态的估计更加精准。

In this paper, a quaternion extended Kalman filter (QEKF) data fusion algorithm is proposed, which is applied to the attitude estimation of quadrotor. Firstly, quaternion is used to represent the coordinate rotation relation in the attitude heading reference system (AHRS), to obtain the attitude information of quadrotor. Secondly, using the extended Kalman filter data fusion algorithm, the 9-axis original data of gyroscope, accelerometer and magnetometer sensors in MPU9250 are combined to filter out the interference noise of the sensor and track the true attitude of the quadrotor. Finally, QEKF is proved to be more accurate in estimating the attitude by comparing it with complementary filter and gradient descent in MATLAB.

TP212.9;TP242;TP391.9

国家自然科学基金资助项目;四川省自然科学基金重点资助项目;中国科学院自动化研究所复杂系统管理与控制国家重点实验室开放基金资助项目;人工智能四川省重点实验室开放基金资助项目;校级研究生创新基金资助项目