You can place our sensors in your field when the insects arrive, the sensors can figure out what type of insect it is. If it is a beneficial or neutral insect, it does nothing. However, if it is a bad bug aka pest, it sends the information to the cloud.

Implementing AI Data for Pest Management
Implementing AI Data for Pest Management

Q&A with Dr. Eamonn Keogh co-founder | Farmsense

Tell us about yourself and your role with Farmsense.io

I am Dr. Eamonn Keogh, I have been a professor for twenty years. I work in the area of data mining and artificial intelligence. Many researchers apply those tools to study humans, how they shop, how they vote, how they age etc. I wanted to do something different, and I always had an interest in natural history, so about ten years ago I started an academic project in what I call “Computational Entomology”, using the power of modern computers to help understand insects.

About 4 years ago, I realized that my work has some commercial possibilities. However, I was reluctant to start a company because; I am not a hardware person and I am not an entomologist. By great fortune, I met two people that became co-founders. Shailendra Singh got his Ph.D. in computer networking with one of my colleagues at UCR, and he is just a genius at hardware, sensors, embedded systems, etc. At about the same time, I met Dr. Leslie Hickle, she is an entomologist (also a Ph.D. from UCR), but in addition, she had a ton of businesses and commercialization experience. I realized that with this core team, we could do something special.  We've also added seasoned advisors through our own contacts and UC Riverside's Entrepreneur-in-Residence members:  CFO, Field Entomologist, Marketing and Sales, Manufacturing, and Agriculture.
 

What is your Smart Pest Monitoring System and what makes it unique?

For every crop humans grow, there is at least one insect pest that wants to eat it! For example, for almonds, the main pest is a moth called Navel OrangeWorm. Growers can typically control these pests, with some combination of interventions, such as pesticides, or biological controls (releasing natural predators). However, in order to do this, they need to understand what insects are in their field (typically down to the sex and species level) and when they first start to arrive.

The way that they do this now has not changed in fifty years. They put mechanical sticky traps in the field, every few weeks they examine the traps, and count the bugs. The problem is that this has a huge cost in labor, and the information they get is stale, up to two weeks old.

FarmSense solves this problem. You can place our sensors in your field when the insects arrive, the sensors can figure out what type of insect it is. If it is a beneficial or neutral insect, it does nothing. However, if it is a bad bug aka pest, it sends the information to the cloud. So, from a farmer's point of view, she might get an occasional automatic text that says something like “Seven Navel OrangeWorm males counted in trap A17 in the last four hours”. With that kind of detailed information, she can plan pinpoint interventions that are cheaper, yet more effective.  
 

How is AI being integrated into pest management?

We look at insects, but there are many companies looking at other things that affect crops, such as temperature, water, weeds, etc. There are approaches that use satellite data or images from drones, and there are some prototype robots that roam the field looking for weeds (and in some cases, removing them!). In general, there has been an explosion of AI and robotics in farming in recent years, at least at the prototype phase.

There are about a half dozen companies that work directly with insect pests like us. However, our unique advantage is that we can accurately tell when and which insects have arrived. Our competitors only know some insect has arrived. But insects can be pests, or neutral, or beneficial, so you really need to know what bugs you have.   Knowing when they first arrive is critical for understanding how the population will develop and when the damaging juvenile forms will be developing from the eggs the flying adults are laying. The larvae are the stage that the farmers are treating with pesticides, but if you know more precisely when the youngest (smallest) stage is there, you have treatment options other than toxic pesticides.

 

What kind of data is being used from Farmsense and what are the decisions farmers can make from the data?

We provide the growers with a detailed “heatmap” that tells them what insects are in their field, in real-time (we also provide some localized temperature and humidity data). They can use this information to make better decisions about interventions. In particular, where to intervene, and when to intervene.

 

Do you have a real-world example you can share with us on a farm having success with your product?

We are generally very protective of our customers' privacy. But I got permission to share this story. Growers sometimes use “puffers”, which release pheromones that confuse the males, ensuring no mating takes place. We had a customer that normally used his puffer from dusk till dawn. But when he saw his data from our sensors, he realized the moths only arrived in a narrow window from 3am to 5am. So he was able to reduce his pheromone cost significantly.  Each species has a preferred time to fly; it's affected by the weather and the stage the crop is in.  We are the first to be able to detect this in real-time through the entire growing season of a crop.

 

How can computer science add value and streamline farm operations to generate growth?

There are many opportunities for computer science to help growers.  Some of it is at the back-end, computers are used to design new seeds, fertilizers, pesticides. Just as Uber lets many people “share” a car, there are Uber-like systems to allow growers to share expensive but rarely used farm equipment. However, I think the biggest bang for the buck is to see farming as an information-driven domain and optimize everything the grower does, from tractor routes to irrigation scheduling, to the timing and placement of pesticide applications.  

Several Crop Management systems have been developed primarily to aid farmers in tilling and planting but none of them have the capability to automatically monitor insect pests and predict the economic impact of treatment or no treatment decisions.  We will be the first to implement a tool to help farmers make this decision using real-time data and a dynamic economic threshold model that is constantly updated in our database for each insect pest and crop.

 

What do you see as the biggest hurdle for expanding AI for Pest Monitoring and how are you overcoming this?

One hurdle is making things affordable. Many people have great ideas but don’t realize how thin the grower’s profit margins are. A wonderful device that costs $5,000 per acre is just not likely to have any impact on agriculture. We spent two years doing nothing but reducing the cost and robustness of our system. We wanted it to be so affordable that if one occasionally gets stolen, or driven over by farm equipment, it is not a big deal.  

Another challenge is attracting talent. Silicon Valley is a magnet for AI talents, and there are thousands of PhDs there that do nothing but use AI to make you slightly more likely to click on some advertisement. It can be hard to compete with the Googles, Facebooks, Amazons of the world. However, some talented AI folks do “get it”. 

At the end of the day, we help overworked farmers grow more food for humankind, which is a fun and noble thing to do.  Our FlightSensor platform helps accelerate the reality of Precision Agriculture which creates sustainable farming ecosystems and insures our future Food Security.

 

 

About Dr. Eamonn Keogh
Dr. Keogh is FarmSense’s Co-founder and Professor of Computer Science at UC Riverside. His career in research and industry has focused on data mining, machine learning, and information retrieval, specializing in techniques for solving similarity and indexing problems in time-series datasets. He is one of only three people in the world to have at least 20 papers in each of the top data mining conferences (SIGKDD, ICDM, SDM). He is listed by Microsoft Academic Search as the 5th most influential data miner in the world.

 

The content & opinions in this article are the author’s and do not necessarily represent the views of AgriTechTomorrow

Comments (0)

This post does not have any comments. Be the first to leave a comment below.


Post A Comment

You must be logged in before you can post a comment. Login now.

Featured Product

XLamp® XP-L Photo Red LEDs – Top Tier Efficiency for Next-Gen Luminaires

XLamp® XP-L Photo Red LEDs - Top Tier Efficiency for Next-Gen Luminaires

XLamp® XP-L Photo Red LEDs deliver a higher class of efficiency for horticulture luminaires. At 83.5% typical WPE at 700 mA, 25°C, the XP-L Photo Red LEDs offer 6% greater efficiency than XLamp XP-G3 Photo Red LEDs. Featuring S Line technology, XP-L Photo Red LEDs offer the best system reliability through switching and dimming cycles with excellent corrosion resistance.