SBIR Phase I: Machine Learning Driven Synthetic Sensor for Plant Water Stress

SBIR 第一阶段：机器学习驱动的植物水分胁迫合成传感器

• 批准号: 1843254
• 负责人: Adam Koeppel
• 金额: $ 22.5万
• 依托单位: Aquasys LLC
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-02-01 至 2020-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1843254&HistoricalAwards=false
• 关键词: SBIR; Phase; Machine; Learning; Driven

The broader impact/commercial potential of this Small Business Innovation Research (SBIR) project is to increase yields and save irrigation water and costs for farmers of perennial crops and high intensity annual crops. In order to maximize yields while simultaneously minimizing irrigation water usage, farmers need to understand, predict, and manage plant water stress, which is the yield reducing stress plants undergo when they struggle to draw water from the soil. If successful, this project will demonstrate to farmers that low-cost measurement and forecasting of plant water stress is possible, and that Machine Learning (ML)-based recommendations for irrigation scheduling reduce water and energy usage, eliminate plant water stress, and increase crop yields. Research and Development activities completed will enhance the understanding of the relationship of plant water stress to microenvironmental conditions and how plant water stress and microenvironmental conditions can be accurately measured and forecast with low cost sensing capabilities.This SBIR Phase I project proposes to perform research and development to validate the concept of a low-cost synthetic sensor for plant water stress. The objective of this project is to field test the low-cost sensor arrays, compare the output to data from high-cost scientific grade sensors, and utilize the data gathered during the field test to train ML models. These ML models will perform sensor fusion on the data from the low-cost sensor arrays and output a plant water stress measurement. These ML models can then forecast future plant water stress, which will be compared to actual field measurements for accuracy assessment and refinement. Once refined, the ML models will determine the minimal amount of irrigation water necessary to mitigate the predicted plant water stress. This optimized amount of irrigation water will be allocated into a recommend irrigation schedule for farmers to review and implement. The relevance and ease of use of the recommended irrigation schedule will then be assessed by farmers.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

这个小企业创新研究（SBIR）项目的更广泛的影响/商业潜力是提高产量，节省灌溉用水和多年生作物和高密度一年生作物农民的成本。 为了最大限度地提高产量，同时最大限度地减少灌溉用水量，农民需要了解，预测和管理植物水分胁迫，这是植物在努力从土壤中汲取水分时所经历的产量降低应力。 如果成功，该项目将向农民证明，低成本测量和预测植物水分胁迫是可能的，基于机器学习（ML）的灌溉调度建议可以减少水和能源的使用，消除植物水分胁迫，提高作物产量。 研究开发活动的完成将提高对植物水分胁迫与微环境条件之间关系的理解，以及如何通过低成本的传感能力准确测量和预测植物水分胁迫和微环境条件。SBIR第一阶段项目计划进行研究开发，以验证低成本的植物水分胁迫综合传感器的概念。 该项目的目标是现场测试低成本传感器阵列，将输出与高成本科学级传感器的数据进行比较，并利用现场测试期间收集的数据来训练ML模型。这些ML模型将对来自低成本传感器阵列的数据进行传感器融合，并输出植物水分胁迫测量结果。 然后，这些ML模型可以预测未来的植物水分胁迫，并将其与实际现场测量进行比较，以进行准确性评估和改进。 一旦完善，ML模型将确定减轻预测的植物水分胁迫所需的最小灌溉水量。 这一优化的灌溉水量将被分配到一个推荐的灌溉计划中，供农民审查和实施。 该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。