基于大数据监测的植保无人机动作决策研究

The Study on Action Decision-Making System for Plant Protection UAV Based on Big Data Monitoring

为了实现山区果树喷洒农药自动化,基于大数据监测,采用卫星导航与图像识别技术,完成了植保机动作动作决策,实现了快速转场与喷药精确飞行。采用GNSS卫星导航技术,设定山区果园喷药开始位置及其航迹点,实现快速转场。利用图像识别处理技术,基于过绿灰度化方法,建立图像灰度计算式,采用大津法实现图像分割;对图像进行中值滤波和膨胀腐蚀填充方法,清晰树冠边界;提取树冠边界点坐标,采用二次多项式拟合,计算航迹线及航向角,并建立卫星导航与图像识别导航转换规则。测试结果表明:植保机航迹与测试点吻合,最大误差为0.45m。

Inorder to achieving spraying automation for fruit trees in mountain area,based on big data monitoring,the action decision-making system of plant protection UAV was achieved to quickly transfer and finish precision flying by satellite navigation and image recognition.Spraying start point and track point were set,then quick transition was achieved by GNSS satellite navigation.The image recognition was used to achieve precision flying.Gray algorithm was achieved based on over green gray algorithm.Crown boundary was found clearly by median filtering and expansion corrosion filling.Pixel coordinates was taken,and then track line and heading angle was calculated by conic fitting.The rule for satellite navigation and image recognition was achieved.The image recognition navigation was tested,the result showed that the track line was identical with test point,and maximum error was 0.45 m.