Development of Levee Weeding Robot

现在水稻种植的大部分工作都是通过拖拉机、水稻种植机、水稻联合收割机等骑式农业机械来进行的。水稻种植中的堤坝除草是日本剩下的严峻工作之一。本研究工作的目的是开发自主工作的堤防除草机器人。本文的研究成果总结如下：1)除草机器人主体选用履带式车辆。在车辆侧面安装了近红外光束型传感器，用于检测堤防形状；在车辆侧面安装了光纤陀螺仪传感器，用于检测车辆航向角。割草机安装在车前。2)提出了导航策略。利用车辆前方虚拟焦点与参考线（即防洪堤中心线）之间的偏差来确定车辆的转弯角度。车辆可反复使用转弯和急转弯。仿真结果与实验结果基本一致，证明了所提出的策略。3)采用机器视觉分析方法对堤防边缘进行检测。从数字摄像机采集图像，进行滤波处理，用HSI变换得到二值图像，用线性回归线或霍夫变换检测堤防线。霍夫变换可以很好地检测复杂图像中的堤线，即使在二值图像中存在一些海峡点。GPS除在水田周围高大树木附近使用外，对车辆位置检测效果良好。由于离合器的调整延迟，自动割草程序无法进行。然后项目将持续进行，以找出机器人对堤防除草的适应性。

Most of the works for rice farming is now carried out by riding type agricultural machines such as tractors, rice planting machines, rice combines and so on. The levee weeding in rice farming is one of the remaining severe work in Japan. The purpose of this research work is carried out to develop levee weeding robot that works autonomously.The results of this research work are summarized as follows :1) The crawler type vehicle was selected for main body of the weeding robot. The near infrared beam type sensors are installed on the side of the vehicle to detect the shape of the levee and fiber optical gyroscope sensor was also installed to detect vehicle heading angle. The mower was installed in front of the vehicle.2) The navigation strategy was proposed. The turning angle of the vehicle was decided by using the deviation between the virtual focusing point in front of the vehide and referenced line that is the center line of the levee. The vehicle can be run to use turning and strait running repeatedly. The simulated and experimental results are almost the same by proving the proposed strategy.3) The machine vision analysis was applied to detect the edge of the levee. The image was taken from degital video camera and treated to filtering, then HSI transform was made to have binary image and linear regression line or Hough transform was made to detect the levee line. Hough transform was very effective to detect the levee line from complicated images even if there are a few strait points on the binary images.The GPS was effective for detecting the vehicle position except to use near tall trees surrounding the paddy field. The autonomous mowing procedure could not be in the process because of the delay of adjustment about the clutch. Then the project will be continuously carried out to find out the adaptability of the robot on the levee weeding.