Aerial robotics sampling for agricultural scouting

自2004年以来，粮食价格一直在上涨，原因不仅是需求增加，还包括化肥/农药成本上升、油价上涨、劳动力短缺和极端天气。为了优化这些作物的产量和健康状况，目前的技术包括定期“侦察”，目视检查作物的生长阶段、营养缺乏、伤害、疾病和真菌，但更重要的是，允许对组织、土壤和昆虫进行取样。这样可以及早发现营养缺乏、真菌感染或虫害。侦察可以提高生产力，降低运营成本，减少化肥和农药对环境的影响。不幸的是，这些技术需要大量熟练工人的季节性帮助。因此，这些工作在最抢手的职位中排名前二，也是最难的。因此，农业发展的主要障碍是耕地面积大，作物监测效率低。目前，人类侦察兵必须行走大片区域。据估计，“传统”勘探每100英亩需要长达5小时，只有几个数据点，并且存在关键的可重复性问题。随着技术的正确发展，机器人技术可以加快操作速度，帮助填补熟练工人的短缺，并提高采样分辨率和可重复性。快速飞行的无人机已经被用于通过多光谱图像监测植物的压力。然而，一致意见是需要实地侦察，因为通过图像可以检测到的植物压力已经使产量降低到理想的经济水平以下。此外，图像不能清楚地区分不同的植物胁迫源。只有预防性的大规模侦察，加上空间和时间分析，才能提供获得精准农业所承诺的全部好处所需的数据。据预测，机器人侦察将推动下一次农业革命。因此，该提案旨在创建空中采样解决方案，以实现对大面积油田的快速侦察。然而，由于涉及硬件开发的风险和困难，特别是接触式空中机器人，目前很少有努力致力于空中农业侦察所需的技术。因此，该计划的短期目标集中在展示新型硬件技术的潜力，其中包括，全自上而下的植物图像，植物组织采样器，土壤采样器和昆虫/真菌采样器。该研究项目将在航空/农业机器人领域培养2名博士，2名MScA和5名coop学员。他们将在顶级会议和科学期刊上发表文章，同时培养行业高需求的技能。除了提高全球粮食产量外，据估计，无人机驱动的农业解决方案的总潜在市场规模为320亿美元。

Since 2004, food prices have been rising due to depleted food stocks resulting not only from increased demand, but also increased fertilizer/pesticide costs, oil prices, labor shortages and extreme weather. To optimize the production and health of these crops, current techniques involve regular "scouting" to visually inspect the crops for growth stage, nutrient deficiencies, injuries, disease, and fungus, but more importantly, to allow for the sampling of tissue, soil and insects. This allows the early detection of nutrient deficiency, of fungus infection or of insect infestations. Scouting leads to better productivity, reduces operating costs, and reduces the environmental impacts of fertilizers and pesticides. Unfortunately, these techniques require the seasonal help of a large number of skilled workers. As a result, these jobs are ranked in the top 2 in-demand positions and hardest. Accordingly, it is claimed that the main obstacle to farming is the large area of farmed land and low efficiency in crop monitoring. Currently, human scouts have to walk large areas. It is estimated that "traditional" scouting requires up to 5h per 100 acres for only a few data points, with critical repeatability issues. With the right technological developments, robotics could speed up operations, help fill the shortage of skilled workers, and increase sampling resolution and repeatability. Fast flying drones are already being used to monitor plant stress through multispectral imagery. However, the consensus is that on-the-ground scouting is required as plant stress than can be detected by imagery will have already decreased production below desirable economic levels. Furthermore, imagery does not clearly differentiate between the different plant stressors. Only preventive large-scale scouting, together with spatial and temporal analysis, can provide the data needed to reap the full benefits promised by precision agriculture. It is predicted that robotics scouting will drive the next agriculture revolution. This proposal thus aims at creating aerial sampling solutions that will enable fast scouting of large fields. However, due to the risks and difficulties involved with hardware development, particularly of in-contact aerial robotics, few efforts are currently devoted to the technologies required for aerial agricultural scouting. As such, the short-term objectives of this program focus on demonstrating the potential of novel hardware technologies, which include, full top-to-bottom plant imagery, plant tissues samplers, soil samplers and insect/fungus samplers. This research program will contribute to training 2 PhD, 2 MScA and 5 coop trainees in the field of aerial/agricultural robotics. They will publish in top ranked conferences and scientific journals while developing skills in high demand by the industry. Beside improving the worldwide food production, it is estimated that the total addressable market of drone powered solutions in agriculture is $32B.