Developing robust sensors for IoT applications in precision agriculture

该研究计划的长期目标是增加传感器的多样性，以及在不同空间和时间的农业场景下运行的可靠性。未来，偏远农场的传感器将发出数据进行管理，机器将变得更加自主。短期的具体目标是：- 发现电磁波和叶面营养素之间的关系，以开发用于作物健康参数的新型非侵入式传感器，-设计机器视觉系统，以区分农业材料和物体，包括植物，石头，土壤，作物等，以在农业机械中实现自主操作或决策支持系统，- 为传感器设计电源和连接方案，使它们能够在农场条件下工作;发明混合能源收割机，可以操作新开发的传感器，这些传感器基于农场中丰富的能源，如机器振动，太阳光和风能;并在传感器中集成低功耗数据传输组件。第一个目标取决于研究电磁波长的基本原理及其与生物材料的相互作用。开发光谱传感器的方法需要通过实验室测试创建数据集，并实施统计建模以开发估计模型。第二个需要图像处理和机器学习的应用程序来操纵图像数据集中捕获的光谱数据。机器视觉系统的精度将受到诸如光源和摄像头等组件的设计的影响。第三个依赖于在农业背景下寻找可再生能源，并发明混合能源收集方法，以保证持续的电力供应。这将通过在将机器视觉系统集成到能量采集器之前降低机器视觉系统连接的功率要求来实现。 新型传感器将解决远程和可靠数据收集的挑战。这是在食品供应链和机器自动化的详细管理方面加强数字农业的必要步骤。在加拿大，大多数农场都很偏远，这些传感器的应用将减少数据收集和维护的旅行需求。在与麦凯恩食品和马铃薯新玩法公司合作的同时，进行这项研究计划的能力为测试传感器提供了必要的场地和设施。此外，由于麦凯恩食品公司在全球马铃薯加工业中的领先地位，它增加了在当地和全球范围内快速适应商业化的机会。HQP培训将侧重于教育和研究方法，以及通过原型设计和测试新型传感器和研究输出方法的项目管理技能。这将使工程专业的学生在职业生涯中为学术和工业工作环境做好准备。

The long-term objective of this research program is to increase diversity of sensors, and reliability to operate under different spatial and temporal agricultural scenarios. In the future, sensors in remote farms shall emit data for management; and machines will become more autonomous. On the short term, the specific objectives are: - To discover the relationship between electromagnetic waves and foliar nutrients to develop novel non-invasive sensors for crop health parameters, - To design machine vision systems, to distinguish between farm materials and objects including plants, stones, soil, crop, etc. to be implemented within farm machinery for autonomous operations or decision-support systems, - To design power and connectivity schemes for sensors to enable them working in farm conditions by; inventing hybrid energy harvester that can operate newly developed sensors based on abundant energy in farm such as machine vibration, solar light, and wind; and incorporating low-power data transmission components in the sensors. The first objective depends on studying the fundamentals of electromagnetic wavelengths and their interactions with biological materials. The methodology of developing spectral sensors requires creating datasets through lab testing and implementing statistical modelling to develop estimation models. The second requires applications of image processing and machine learning to manipulate the spectral data captured in image datasets. The accuracy of the machine vision system will be influenced by the design of components such as light and camera. The third relies on finding renewable energy within the agricultural context and inventing hybrid energy harvesting method to guarantee continuous power supply. This will be achieved by reducing the power requirement of the machine vision system connectivity before integrating it to the energy harvester. The novel sensors will address challenges of remote and reliable data collection. This is a necessary step towards enhancing digital agriculture in terms of detailed management of the food supply chain and machine automation. In Canada, where most of farms are remote, the application of these sensors will reduce travelling need for data collection and maintenance. The ability to conduct this research program while partnering with McCain Foods and Potatoes New Brunswick provides field and facility accessibility necessary to test sensors. Also, it increases the chance of fast adaptation for commercialization on local and global scales because of McCain Foods' leading position in potato processing industry globally. The HQP training will focus on educational and research methodologies, and project management skills through prototyping and testing novel sensors and methodologies of the research output. This will prepare engineering students for both academic and industrial work environments as they move forward in their careers.