Grand Challenges is a family of initiatives fostering innovation to solve key global health and development problems. Each initiative is an experiment in the use of challenges to focus innovation on making an impact. Individual challenges address some of the same problems, but from differing perspectives.
Peter Wagstaff of Self Help Africa in Ireland will build an advanced machine learning algorithm that automatically analyzes high-resolution satellite images for near real-time, low-cost detection of crop pests and diseases across wide, varied landscapes. Current detection methods are either resource- or cost-intensive and limited in their ability to provide up-to-date information across large and complex geographic areas. Crop pests and diseases can alter leaf color and expose soil, which can be detected by very high-resolution satellite imaging. They will combine satellite images provided by their partner with field data on the fall armyworm crop pest collected by their project team over 18 months on smallholder plots in the Balaka district in Malawi. These data will be used to train an algorithm to detect pests and diseases. They will use cloud-based workflows to enable computationally intensive processing of large quantities of high-resolution images in near real-time. The accuracy of the algorithm will be evaluated by an independent field survey. Note: This grant is funded by the Foundation for Food and Agriculture Research (FFAR).
Hanseup Kim of the University of Utah in the U.S. will develop small, ultra-low power, chemical sensors that can be distributed around farms to help detect crop diseases in low-resource settings. Plants under attack from pests and diseases release low levels of volatile organic compounds that could be used as an early warning system to reduce crop losses, which can be substantial. They will design chemical sensors that trigger a change in electrical conductivity when they bind a target compound to minimize energy consumption so that they can be operated over the eight-month farming season in low-resource settings. The sensors will first be developed to bind trace levels of hexenol, hexenal, or indole, which are released from damaged maize and sorghum. They will optimize sensitivity by testing different sensor materials and correlate compound detection with different types and stages of crop damage. They will also evaluate wireless monitoring of multiple sensors distributed around a small plot of crops ready for scaling up to future in-field testing on these and other crops. Note: This grant is funded by the Foundation for Food and Agriculture Research (FFAR).
William Kunin of the University of Leeds in the United Kingdom will develop methods to monitor agricultural pest outbreaks in Africa using data from dual-polarization weather radar. Pest infestation is responsible for up to 50% of pre-harvest crop loss in Central Africa, and control depends on the ability to monitor local pest outbreaks and movement over large areas – a difficult and expensive task. Sophisticated dual polarization Doppler weather radar is designed to detect airborne objects like rain and hail. However, because it is sensitive to size and shape, it can also be used to detect specific types of insects, including crop pests, and algorithms exist to separate the different signals. They will use micro-CT scans to create three-dimensional models of pests common to Rwandan crops and conduct simulations to predict radar data patterns for swarms of particular pests at varying densities. Through collaboration with the Rwandan meteorological service, they will analyze radar data to identify pest outbreaks and their movements across the country. Once established, the methods will be applied across Africa to provide a low-cost, advanced warning system for crop protection.
Molly Brown of Syngenta Foundation for Sustainable Agriculture in the U.S. will develop an early warning system for crop diseases for rural farmers in Africa by gathering data using existing infrastructure and mobile tools from commercial partners and applying machine learning pest models. Insect pests cause almost half of crop losses in Africa each year, which impacts both food supply and the economy. An effective disease and pest surveillance system provides early warning to minimize crop loss but is often unavailable to rural farmers in poor countries who lack tools for sufficient pest data collection. To address this, they propose to leverage the existing infrastructure, including demonstration farms, retailers, and mobile tools, of a commercial partner to gather the required data, such as field photographs, data on pest prevalence, and satellite weather data. Mobile tools will be available on computer and Android devices with off-line modes. They will pilot test their approach in Zambia focusing on damage to maize crops caused by the fall armyworm. Real-time data will be uploaded to a freely available, public portal with information on geography, pest populations, and affected crops.
Ritvik Sahajpal of the University of Maryland College Park in the U.S. will develop an early warning system for low-income countries that predicts the threat to crops from pests and diseases by combining machine learning and crop pest modelling with freely available earth observation data. Existing monitoring systems allow farmers to share data on pest incidence to ensure the timely and limited use of treatments. This maximizes crop yield while minimizing cost and environmental damage. While effective, these systems are expensive and logistically challenging in low-resource settings, particularly as they require widespread coverage and monitoring of a range of pests and diseases. They have designed a new, low-cost early warning system to automatically predict a variety of different crop threats using freely available data and will first test it on maize and sorghum crops in Tanzania. They will use an ensemble-based machine learning model to estimate crop losses two to three months before harvest using earth observation datasets including vegetation indices, temperature, and soil moisture. They will then determine how much of these pre-harvest losses are caused by crop pests such as fungi and insects using the Environmental Policy Integrated Climate Model, which simulates their impact on plant health including growth. They plan to integrate their warning system with existing agricultural monitoring networks to improve accuracy. Note: This grant is funded by the Foundation for Food and Agriculture Research (FFAR).
Amit Lal of Geegah LLC in the U.S. will develop battery-powered ultrasonic imagers to collect and wirelessly transmit high-resolution images of soil and airborne pests for the early detection of crop threats across large farming areas in rural Africa. Crop losses due to pest infestation negatively impact both food security and local economies. Damage caused by nematodes is particularly difficult to detect because the symptoms visible above ground are not unique and are often incorrectly attributed to deficiencies in soil nutrients or moisture. Currently the only way to test for nematodes is through root and soil samples taken after harvest, when it is too late to respond to an infestation. They will integrate complementary metal-oxide semiconductors (CMOS) with GHz ultrasonic imagers to detect nematodes and survey soil properties over large areas to detect infestations before crop damage occurs and transmit the data in real time. They will optimize the sensor technology in a controlled laboratory setting to maximize sensitivity and specificity and minimize power consumption and then transition to a farm setting for incorporating data transmission via the radio frequency wireless network.
Firat Guder and Tony Cass of Imperial College London in the United Kingdom along with George Mahuku and James Legg at the International Institute of Tropical Agriculture in Tanzania will develop a smartphone-based electrochemical lateral flow assay to rapidly diagnose crop viruses in the field and relay the results. Viral crop diseases like maize lethal necrosis and cassava brown streak disease can destroy up to 30% of crops and are the main threat to food security in East Africa. Most diagnostic tests are laboratory-based and slow, hampering efforts to stop the diseases from spreading. They will develop a system that combines cutting edge chemistry and widely-used smartphone technology to quickly test field samples and upload the results to the cloud for immediate sharing with farmers and agricultural partners. Their approach is based on chemical amplification and detection of nucleic acid aptamers: as few as a hundred virus particles can be detected without the need for complex genetics or fragile and expensive antibodies. They will identify aptamers that selectively recognize the viruses, optimize the lateral flow assay in the lab, and then field-test it in Tanzania and the Democratic Republic of Congo. Once validated, they plan to develop a simpler, disposable version and scale-up the technology to other crops and countries.
Melanie Bannister-Tyrrell of Ausvet in Australia will create an SMS-based communication system for farmers in Kenya to anonymously report crop disease and pest infestations and generate surveillance data to minimize crop loss. Pest infestation and disease cause substantial crop losses each year. In many low-income countries, farmers do not report disease to local agricultural authorities because they fear their crops will be destroyed without compensation. Yet information on the presence and spread of pests is needed to inform decisions on planting and control measures. They will optimize the design of an SMS interface for anonymous reporting using SMS codes for diseases and pests. The system will be piloted in one county, leveraging existing users and community volunteers to recruit new users. Reports will be used to estimate the local prevalence of pests and disease using modified small-area statistics. If a credible threat is detected, all users will be alerted. The system will improve connection and trust between farmers and agricultural extension workers and improve surveillance.
Galaletsang Tsontswane of Congretype in South Africa will design low-cost, solar-powered insect detection traps equipped with wireless sensors to capture images of insects and transmit them to a central control point to improve rural crop surveillance in developing countries. Crop loss due to pest infestation negatively impacts both food supply and local economies, while rural farmers in developing countries lack resources to monitor crop disease and infestations and are unable to respond before substantial loss occurs. They will design and deploy intelligent insect traps equipped with cameras and test their ability to capture images of trapped insects within a study area and to wirelessly transmit the data to a central location via an existing TV white space network. The data can then be stored and accessed by local farmers and agricultural extension workers, allowing them to respond more quickly to crop threats than is currently possible based on data from laborious, human-based monitoring systems.
Joyce Nakatumba-Nabende of Makerere University in Uganda will use artificial intelligence to mine data from local village radio stations to generate timely data on crop pests and disease in sub-Saharan Africa. Crop loss due to pests and disease threatens the economic survival of smallholder farmers, and access to surveillance data is critically important yet often unaffordable. Local radio shows are a powerful source of information flow in rural African villages: they cover topics including politics, policy, climate, and social circumstances, in addition to crop concerns. Collectively, this information provides a holistic representation of current events in these communities. They will analyze local broadcasts to generate crop surveillance data that is linked to the local community situation. Radio content will be collected at low cost through a collaboration with Pulse Labs Kampala, and they will build artificial intelligence models based on deep neural networks and keyword identification to mine the data. The results will be combined with photographs of diseased crops provided by local farmers and used to train machine learning models to ultimately extract radio information in multiple languages and with diverse accents. This project will provide near real-time crop surveillance data and allow for timely responses to threats.