基于梯度下降法和自适应参数相结合的姿态解算方法

Attitude Estimation Based on Gradient Descent Algorithmand Adaptive Parameters

为了解决四旋翼无人机在姿态解算时的高精度和实时性问题,提出了一种基于参数自适应的梯度下降法和互补滤波相结合的多传感器数据融合算法。该算法采用四元数表示姿态信息,利用梯度下降法对磁力计和加速度计数据进行预处理,并根据陀螺仪输出的角速度和外部加速度大小自适应选择梯度下降参数β,再将其和陀螺数据更新后的四元数进行互补滤波用于补偿陀螺的累积误差,解算出三个姿态角。最后设计仿真与实验分析。实验结果表明,相对于传统的梯度下降法和互补滤波法,该算法姿态估计误差小且具有更好的静态和动态性能。

In order to improve the real-time and high-precision problems of quadrotor UAV during attitude estimation,a multi-sensor data fusion algorithm based on adaptive parameter gradient descent and complementary filtering is proposed.The algorithm uses quaternions to represent attitude information.The gradient descent method is used to preprocess the magnetometer and acceleration data,and adaptively select the gradient descent parameter according to the angular velocity output by the gyroscope,and then it is complementary filtered with the updated quaternion of the gyroscope data to compensate for the integra error of the gyroscope.Finally,simulation and experimental tests show that the algorithm has a small measurement error,and has better static and dynamic performance than the traditional gradient descent algorithm.