基于改进 ESKF 的植保无人机时延位姿补偿算法

为解决全球导航卫星系统和惯性测量单元融合时间不同步问题,提高植保无人机位姿估计精度,本文根据植保无人机 大惯性、强振动的特性提出一种基于改进误差状态卡尔曼的时延位姿补偿算法。 首先对名义状态变量线性预测,引入渐消因子 提高强振动环境下的系统稳定性;接着采用互补滤波对角速度补偿,对姿态误差状态变量修正;最后结合测量的延迟时间,使用 互补滤波外推数据,提高大惯性特性下的速度位置精度。 实验结果表明,相较于误差状态卡尔曼算法,横滚角和俯仰角均方根 误差减少 0. 266 9°和 0. 241 4°,偏航角均方根误差减少 0. 076 4°;正常航迹植保作业下,东北天方向速度均方根误差减少 0. 210 5、0. 184 9、0. 238 8 m/ s;东北天方向位置均方根误差分别减少 0. 21、0. 19、0. 23 m,有效提高位姿估计精度。

Time delay and attitude compensation algorithm for plant protection UAV based on the improved ESKF

To solve the problem of global navigation satellite systems and inertial measurement unit fusion time asynchronous and improve the accuracy of pose estimation of plant protection UAV, this article proposes a delay pose compensation algorithm based on the improved error state Kalman filter by using the characteristics of large inertia and strong vibration of plant protection UAV. Firstly, the nominal state variables are linearly predicted, and a fading factor is introduced to improve the system stability in strong vibration environments. Then, complementary filtering is used to compensate for diagonal velocity and correct the attitude error state variables. Finally, combined with the delay time measured, complementary filtering is used to extrapolate the data and improve the velocity and position accuracy under high inertia characteristics. Experimental results show that, compared with the error state Kalman filter algorithm, the root mean square error of roll angle and pitch angle is reduced by 0. 266 9° and 0. 241 4°, and the root mean square error of yaw angle is reduced by 0. 076 4°. Under normal track plant protection operation, the root mean square error of velocity in the northeast sky direction values are decreased by 0. 210 5, 0. 184 9, and 0. 238 8 m/ s. The root mean square errors of the northeast celestial position are reduced by 0. 21, 0. 19, and 0. 23 m, respectively. The algorithm effectively improves the accuracy of pose estimation.