Detecting asymptomatic disease through drones and AI

Much like humans, plants can spread diseases asymptomatically, making plant diseases difficult to detect and control.  Through his research, Dr. Enrico Bonello is developing drone-mounted sensors coupled with AI technology to detect infections and infestations in plants for more effective disease and pest management. View Halo Profile >>

Tell us about your research

I am a plant molecular and chemical pathologist/ecologist. My mission is to translate basic science discoveries we make on established and emerging diseases and insect pests into applications. These include tools to screen plants for resistance and to accurately detect plant disease or insect attack in a timely manner. In this project, we want to develop the use of near infrared spectroscopy-based (NIR) sensors mounted on small unmanned aerial vehicles (sUAV, AKA drones), coupled with artificial intelligence approaches, to detect asymptomatic plants that are infected or under attack by “reading” their chemical fingerprints associated with physiological changes due to infection.

My mission is to translate basic science discoveries we make on established and emerging diseases and insect pests into applications.

Can you explain that to a non-scientist?

Imagine yourself being infected with the COVID-19 virus but being asymptomatic. It’s the same for plants, except plants cannot move when in the field. These asymptomatic plants are located outside the boundary of the symptomatic area, which makes delineating the actual extent of infection very hard. By “reading” the plants to detect chemical signatures of physiological changes due to infection, we can rapidly find the true boundary of an epidemic and treat the plants accordingly.

By “reading” the plants to detect chemical signatures of physiological changes due to infection, we can rapidly find the true boundary of an epidemic and treat the plants accordingly.

Why did you choose this area of research? 

Plant diseases and insect pests are among the most significant limiting factors in agricultural productivity and forest ecosystem stability. One of the key obstacles to proper management of these problems is our inability to quickly diagnose a plant as infected or under pest attack. Just as important is our inability to do so efficiently at a landscape level. Our research tests the idea that we can detect diseases early and more accurately if we can couple chemical fingerprints to phenotype plants with the capacity to deploy the sensors using sUAV over entire fields.

How could your Grants4Ag project someday impact #healthforall #hungerfornone?

The ability to detect infected, asymptomatic plants will provide a head start on disease and insect pest management in different ways. For example, better delineation of infection zone boundaries will allow a more targeted approach to treat or destroy those plants that can act as “silent spreaders”. It will also allow the detection of plants genetically engineered for resistance to pests (e.g. with Bt genes) when that resistance breaks down. In that case, such plants can be treated with insecticides in a targeted manner and/or replaced with different sources of resistance. In all these cases, management can be improved via reduction of fungicides and pesticides, resulting in more healthful and abundant agricultural production.