Agricultural robot platform based on decision-making logic by proficient in farming

研究表明，针对农业机器人遥操作难的问题，提出了车辆控制机构时延补偿方法和数据通信时滞补偿方法。此外，还开发了使用卫星图像地图的用户界面，以向操作员指示准确的现场信息，以确保操作安全。现场测试表明了所开发方法的有效性。具体内容如下：(1)提出了HST-杠杆机构的非线性控制方法，并对其死区进行了补偿。(2)现场图像查看器，实现了更安全的操作。1)改进的图像映射方法可以从变形的全景图像中生成平面图像。实验结果表明，在距摄像机中心10m的范围内，最大位置误差为0.6m。2)基于二维和三维A卫星图像地图的用户界面采用Google map API，可以以1赫兹的频率更新车辆的位置。利用OpenSceneGraph开发了3D虚拟田间地图用户界面，有效地减少了机器人与客户端之间的数据包大小，并成功地操作了来自美国的机器人车辆。问卷调查结果显示，农业机器人等信息技术的引入，年轻(非熟练)农民比老年农民(熟练者)更容易介绍，但年轻农民比老年农民更需要更多的信息来进行耕作决策。在拖拉机操作方面，熟练者的眼痕动作比业余驾驶时的动作更尖锐。这些结果表明，农业机器人需要为非熟练农民提供辅助功能，如天气和操作方法、信息系统和操作的推荐视点指示器。

This research shows that compensation methods of time-lag both in the control mechanism of the vehicle and data communication were developed to resolve the difficulty in the teleoperation of the agricultural robots. Besides, user interfaces using satellite image map are developed to indicate precise field information to an operator for safer operations. Filed tests denoted the availability of the developed methods. The details of the methods are described as below.(1) Control method for non-liner systemNon-linearity of HST-lever mechanism was figured out, and the dead zone of it was compensated by the developed method.(2) Field image viewers for safer operations1) The developed image mapping method can generate flat images from distorted omni-vision images. As a result of experiment, maximum position errors are 0.6m within 10m from camera center.2) 2-D and 3-D A satellite image map based user interfacesThe 2-D user interface is used Google map API, and it can update the position of the vehicle by 1Hz. The 3-D virtual field map user interface was developed using the OpenSceneGraph, it was effective to reduce the data packet size between the robot and a client, and the robotic vehicle was operated by it from U.S.A successfully.As results of questionnaire surveys, introduction of IT such as agricultural robot is easier young (unskilled) farmers than aged farmers (proficients), however young farmers tend to need more information for making decision of farming than aged farmers. In terms of tractor operations, movements of proficient's eye marks are sharper than amateur's movements while driving. Those results suggest that agricultural robots need assist functions for unskilled farmers such as weather and operational methods information system and a recommending viewpoint indicator for operations.