NSF Convergence Accelerator Track M: Targeted Insect Sensing and Control

在美国，虫害每年破坏或摧毁价值数十亿美元的食物。控制这种害虫的努力(通常是通过地毯式喷洒刺激性杀虫剂)是昂贵的，而且对环境有害。几十年来，种植者一直在试验自然生物控制，包括释放目标害虫的自然捕食者/寄生虫，以及释放信息素来迷惑雄性并极大地减少它们找到配偶的成功率。然而，这两个想法都需要准确的昆虫监测，知道哪些昆虫在田间的什么地方。获得这类信息的成本和难度意味着生物控制不具有成本效益，因此很少使用。该项目将开发和测试昆虫监测的新方法。该方法大致基于自然界最好的昆虫传感器之一--蝙蝠。蝙蝠可以从相对较远的距离探测和辨别各种昆虫。拟议中的传感器将允许对昆虫进行低至性别、物种和生命阶段的监测，并将提供有关昆虫何时最活跃的关键信息。这一信息对于拟议干预措施的准确时间安排很重要。该项目的更广泛影响将包括降低种植者的劳动力和材料成本，使美国农业更具竞争力。此外，通过减少对刺激性杀虫剂的需求，该项目寻求减少对有益传粉者和环境的破坏。该项目是高度跨学科的，涵盖无线传感、云计算、机器学习、机器人学、昆虫学、统计学、农业经济学和符号化学。最初将使用直接计算指标来衡量进展，例如传感器的信噪比和分类算法的错误率。中期评估将衡量虫害压力和作物损失的减少。衡量成功的最终标准是经济和行为：该系统是否为种植者降低了成本，他们是否愿意采用新技术？该项目将与加州中央山谷的坚果种植者一起衡量这些终极指标。该奖项反映了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。

Insect pests damage or destroy billions of dollars’ worth of food in the USA each year. Efforts to control such pests (often by blanket spraying of harsh pesticides), are expensive and harmful to the environment. For decades, growers have experimented with natural biocontrols, including the release of natural predators/parasites of the targeted pests, and the release of pheromones to confuse the males and drastically reduce their success in finding a mate. However, both ideas require accurate insect surveillance, knowing which insects are where in the field. The cost and difficulty of obtaining such information has meant that biocontrols are not cost effective and are thus rarely used. This project will develop and test novel methods for insect surveillance. The method is based loosely on how one of nature’s best insect sensors, bats. Bats can detect and discriminate between various types of insects from a relatively large distance. The proposed sensors will allow the surveillance of insects down to the level of sex, species, and life stage, and will provide critical information about when the insects are most active. This information is important for the precise timing of the proposed interventions. The project’s broader impacts will include reduced labor and material costs for growers, making US agriculture more competitive. Moreover, by reducing the need for harsh pesticides, the project seeks to reduce damage to beneficial pollinators and the environment.The project is highly interdisciplinary, spanning wireless sensing, cloud computing, machine learning, robotics, entomology, statistics, agricultural economics and semiochemistry. Progress will initially be measured with direct computational metrics, such as the signal-to-noise ratio of the sensor, and the error-rate of the classification algorithms. Intermediate evaluation will measure the insect pest pressure and the reduction in crop loss. The ultimate metrics of success are economic and behavioral: Does the system reduce costs for the grower, and are they willing to adopt the new technology? The project will measure these ultimate metrics with tree nut growers in California central valley.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.