SBIR Phase II: Artificial intelligence powered optical spectrometer technology for farm-level milk testing

该小型企业创新研究(SBIR)第二阶段项目利用移动光谱和机器学习开发了一套牛奶检测系统，以提供快速、自动的牛奶检测。该团队的目标是解决由于缺乏养殖场级别的早期检测技术而造成的每年320亿美元的全球损失。牛乳腺炎是一种乳房疾病。该项目帮助农民及早发现乳房炎，使他们能够提高农场运营效率，减少抗生素的使用，改善动物健康，并减少温室气体排放。全球奶牛场不断扩大的牛群规模、劳动力短缺和不断增长的乳制品消费正在推动全球牲畜监测市场的增长，预计到2030年，全球牲畜监测市场将达到190亿美元。在全球范围内，潜在市场总规模估计为120亿美元。这个二期项目中正在开发的技术将通过将尖端技术带到农场并创造机会来吸引和留住新一代乳品工人，从而实现精准乳品生产。该项目的使命是支持乳制品行业以高效和可持续的方式提供最优质的牛奶。该项目的智力优势包括使用由物理信息机器学习控制的移动光学光谱仪技术，对牛奶成分进行现场、实时和可靠的测试，如体细胞计数、脂肪和蛋白质。将开发一种改进的直列牛奶检测单元的工业设计，该设计是为机器人奶牛场量身定做的。此外，嵌入式采样器原型将在传统奶牛场进行测试，以完全自动化牛奶采样和测试，目标是开发一种适用于大多数客厅配置的通用设备。使用近红外和短波红外芯片将扩大设备的工作波长范围，这不仅将提高脂肪和蛋白质测量的准确性，还将把检测范围扩大到乳糖和牛奶尿素氮等成分。从数据的角度来看，体细胞计数的时间序列测量将与历史牛群水平和个体奶牛水平的数据相结合，如挤奶天数、哺乳天数和产奶量，以建立乳房炎和产奶量的预测模型。最后，荧光等光学信号将被用来确定牛奶中有害病原体的存在，以帮助诊断感染并防止受污染的牛奶进入供应链。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This Small Business Innovation Research (SBIR) Phase II project develops a cow's milk testing system using mobile spectroscopy and machine learning to provide rapid and automatic milk testing. The team aims to address the annual $32 billion global loss from bovine mastitis, an udder disease, due to the lack of farm-level early detection technology. The project helps farmers detect mastitis early, allowing them to increase farm operation efficiency, lessen the use of antibiotics, improve animal health, and reduce greenhouse gas emissions. The expanding herd size of dairy farms, shortage of labor, and rising dairy consumption across the globe are driving growth in the global livestock monitoring market which is expected to reach $19 billion by 2030. Globally, the total addressable market size is estimated at $12 billion. The technology under development in this Phase II project will enable precision dairy production by bringing cutting-edge technology to the farm and creating opportunities to attract and retain a new generation of dairy workers. The project’s mission is to support the dairy industry in delivering the best quality milk in an efficient and sustainable way.The intellectual merit of this project involves on-farm, real-time, and reliable testing of milk components such as somatic cell counts, fat, and protein, using mobile optical spectrometer technology that is controlled by physics-informed machine learning. An improved industrial design of the inline milk testing unit will be developed that is tailored for robotic dairy farms. Additionally, an embedded sampler prototype will be tested in conventional dairy farms to fully automate milk sampling and testing with the goal of developing a universal device that works for most parlor configurations. The operating wavelength range of the devices will be broadened using near-infrared and shortwave-infrared chips, which will not only increase the accuracy of fat and protein measurements but will also expand the testing to components such as lactose and milk urea nitrogen. From the data perspective, time-series measurements of somatic cell counts will be combined with historical herd-level and individual cow-level data such as days-in-milking, lactation, and yield, to build predictive models for mastitis and milk yield. Finally, optical signals such as fluorescence will be used to ascertain the presence of harmful pathogens in milk, to aid in the diagnosis of infections and prevent contaminated milk from entering the supply chain.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.