基于EKF的农机智能体自主导航算法研究

为获得更加准确、全面、实时的农田障碍物信息，提高农业机械智能体自主导航定位的精度，提出一种基于北斗系统和视觉导航的组合定位方法。针对农田环境，选择BDS、视觉CCD为外部传感器，设计一种基于扩展卡尔曼滤波器(EKF)的数据融合算法，该算法融合了BDS和视觉传感器数据，实时定位农机智能体的位置。系统通过对导航角度和行驶进度进行跟踪，完成绝对定位。通过机器视觉图像处理，获取导航基准和作业目标信息，完成相对定位。通过试验验证该算法的有效性，并通过卡尔曼滤波算法(KF)的成果进行对比分析。结果表明：滤波后的路径更平滑，抖动偏差减小，坐标数据比KF滤波结果更稳定、更平滑。此外，距离的平均误差可以从滤波前的0.119 5 m降低到滤波后的0.07 0 m，有效地降低了过程噪声。且位置偏差在±0.1 m以内，精度较高，提升了农机智能体自主导航的定位精度。

Research on autonomous navigation algorithm of agricultural machinery agent based on EKF

In order to obtain more accurate, comprehensive and real time farmland obstacle information and improve the autonomous navigation and positioning accuracy of agricultural machinery agents, a combined positioning method based on Beidou system and visual navigation was proposed. In this paper, a data fusion algorithm based on extended Kalman filter (EKF) is designed, which selects BDS and visual CCD as external sensors. The algorithm integrates BDS and visual sensor data to locate the real time location of the agent. The system completes absolute positioning by tracking navigation angle and driving progress. Through machine vision image processing, navigation reference and target information are obtained to complete relative positioning. The effectiveness of the algorithm is verified by experiments, and the results of Kalman filter (KF) are compared and analyzed. The results show that the filtered path is smoother, the jitter deviation is reduced, and the coordinate data is more stable and smoother than the KF filtering result. In addition, the mean distance error can be reduced from 0.119 5 m before filtering to 0.070 m after filtering, effectively reducing the process noise. And the position deviation is within ±0.1 m, the accuracy is higher, which can improve the positioning accuracy of intelligent agricultural machinery autonomous navigation.