Artificial Intelligence Meets Agriculture at Root AI

Root AIPhoto courtesy of Root AI.

Let us introduce you to Virgo. Think of it as a farmer’s assistant—one that can tell when tomatoes are ripe and nimbly pluck them.

Created by Somerville-founded start-up Root AI, the robot uses artificial intelligence to navigate the complex tasks of indoor, greenhouse, and outdoor farming. And picking tomatoes is just the beginning, according to Co-Founder and CEO Josh Lessing.

What’s your background and how’d you get to the point of founding this company?

I started out my career in food automation, specifically food processing, food packaging [and robotics]. All across the food supply chain, there’s a lot of really tricky problems that need to be solved and can be solved with more modern robotics and artificial intelligence tools. 

That’s what got me into building Root. One of the biggest issues is being able to find people to work in farms—out in the fields, amongst the crops—and then simultaneously being able to optimize those crops so that we can uplift production.

Part of the problem is it’s seasonal work, and that just becomes a logistics issue. If I’m working in Baja, Mexico, and I have a crop harvest that lasts three months, what do I do the rest of my year? These agriculture jobs are highly physical, they’re in harsh conditions, often, and for many kinds of crops the window over which your services are needed is very small. So it’s become a complex labor challenge, all over the entire world.

How do you solve this problem? For us, it’s robotics and artificial intelligence.

What are some of the benefits of having a robot do this work?

There’s a lot that you can get done with a robot. There’s obviously the need of being able to find people to work the farm, but there’s also being able to use what robots are great at to elevate yield. 

One of the things that we can heavily benefit from is more individualized crop care. So if I am overseeing 500,000 tomato plants on a 40-acre farm, and I have deep expertise on how to care for the health of a plant, if I had infinite time I could go to each plant and observe its nutritional health, whether or not there’s insects, whether or not there’s disease, and I could do an amazing job of caring for each and every plant.

That’s not a tractable task for one person, but if you had robotic systems that were farming assistants—and that’s what we’re building, we’re building farming assistants—if you had robotic systems that are driving around the farm, they’re doing harvesting, they’re doing the pruning, a lot of the more physical, dexterous work on the farm, they also have eyes and a brain, so why can’t they be looking at the plant and informing the grower that there are nutritional deficiencies, and simultaneously being able to take immediate action? If your systems know that the grower wants you to do the following under the following set of circumstances, the robots can enact the high-level growing plan, the intent of the grower for those 500,000 plants. So you’re looking at more individualized crop care.

What are some of the things the robot can do?

Right now, we’re focusing very specifically on harvesting for our first product. It is one of the most challenging things in order to be able to oversee at a growing operation. In order to harvest, a robot needs to be able to look at a crop, assess fruits for quality, assess them for ripeness, and then harvest it. Since it’s already visualizing these fruits, it’s in a position to, in the future, be able to say new things about them. Like if it’s looking at a crop, and it knows how many are unripe and what stage of ripeness everything’s at, it can predict future yields. That really helps manage supply chains for growers, for retailers. If it’s looking at the crops, it can also be spotting insects. That’s what we’re trying to work toward, as a holistic platform.

Do humans have any advantage over a robot here?

I would say for a lot of the high-level judgment calls on the farm, that is a great place for people to do work, and then the implementation of those judgments, that’s where the robot comes in, that’s why it’s a robotic assistant. So I wouldn’t say there’s any disadvantages to using a robot, because you wouldn’t build features that would replace things that humans simply just do better. You’re trying, basically, to extend a person’s capacity to get work done by giving them helpers.

Indoor farming is another piece of this equation. What are the benefits of indoor farming?

Incredible levels of efficiency, for one, and on top of that the ability to grow in arbitrary geographies and the ability to grow more sustainably. Because of all the technology that you put into these buildings, you also can do things a lot more sustainably. They’re likely doing [over] 25 times as much fruit production per acre as an outdoor reference point. These types of farms can do up to 90 percent less water use, since all the water goes directly into the plant, and any water that the plant doesn’t need is collected, reprocessed, and recleaned so it can be used again. And then on top of that, because they have this controlled, indoor environment, they can massively drop the pesticide use.

How can this technology help the agriculture industry navigate climate change?

It’s about predictability. If you get to control your growing environment, then you get to control the outcome. If you are using a glass house enclosure or a full enclosure, you have a lot of fine control over the temperature, you have a lot of fine control over water exposure to the crops, humidity, and as a result, you know what amount of production [you’ll have] with a much higher degree of confidence than you would with an outdoor farm today. Looking toward climate change in the future, there’s going to be increasing unpredictability in weather patterns that are going to have really massive effects on outdoor production. 

A big part of what we want to support at Root is making fresh fruits and vegetables much more accessible, and accessibility has to do with having local production, so that people can get high-quality, vine-ripened, fresh produce next-day. This is the problem of today. The problem of tomorrow is unpredictability and weather, which creates fluctuation in supply and pricing.

What stage is Virgo in now?

It’s a late-stage prototype. We’re going to be launching the product early next year, and right now it’s a mobile harvesting platform that has an on-board intelligence, doesn’t require to be connected to the internet, can see tomatoes regardless of their ripeness. And it’s able to assess whether or not they’re ripe enough to be picked, and then be able to pick without damage. The tomatoes that we pick are already sold to grocery stores.

Editor’s note: This interview has been edited and condensed for clarity and conciseness.

This story originally appeared in the Technology & Transportation issue of Scout Somerville, which is available for free at more than 200 locations throughout the city or by subscription.

Like what you’re reading? Consider supporting Scout on Patreon!