Farm Researchers Are Using Military Face-Recognition Software to Inspect Grapes

Why measure sorghum when you can get a robot to do it faster?

Kaleigh Rogers

Kaleigh Rogers

Somewhere in a field in South Carolina, a robot drives slowly through the dry soil between tall, swaying rows of sorghum: a nutrient-rich grain.

The robot scans the stalks of sorghum, photographs them, looking for disease. It uses laser scanning to estimate their height and volume. And, every now and then, it reaches out a robotic arm, grabs hold of one of the stalks, and stabs it with a probe to measure the thickness of the rind.

Welcome to the farm of the future.

This little robo-farmer is just one project developed by FarmView, a multidisciplinary, multi-institution effort to put advanced technology to use on the farm. With an exploding global population, we need to increase food production by 70 percent over the next 40 years if we hope to feed everybody, according to the United Nation's Food and Agriculture Organization. While the impacts of climate change will increase crop yields in some areas, it will introduce new challenges that make it more difficult to grow healthy, abundant crops.

If we hope to meet this demand, we need to find ways to make farming more resource efficient, more productive, and faster. So it makes sense that farm researchers are looking for helpful tech anywhere they can find it. The sorghum robot, for example, uses navigation technology developed for driverless cars, and sensors developed for the military.

"In the US, we've spent a lot of money on defense recently," said George Kantor, senior scientist for FarmView at Carnegie Mellon University. "Algorithms that do face recognition for tracking people through airports and things like that become the algorithms that you can use to recognize grapes. If you just look at where we've been investing in technology research, we're using all of those tools [in agriculture]."

The sorghum-stabbing robot project puts technology to use in the field of crop breeding. Traditionally, crop breeders combine different strains of a crop (there are 40,000 varieties of sorghum) to create many different "children" crops of new strains, which they have to grow to test out. This allows them to breed new versions of the crop that are more resistant to disease, or to drought, or have a higher yield. This is a timely process and tracking all of the different new strains, how they're developing, and what their attributes are, is difficult. But if you can enlist a robot to do it, it speeds the process up significantly, Kantor told me.

The robot grabs a stalk of sorghum. Image: Carnegie Mellon School of Computer Science

Kantor and his team is focused on sorghum not as a food crop (though it is a protein-packed little grain that's popular in some parts of Africa and India) but as a potential biofuel. Kantor said that, right now, sorghum produces 10-15 tons of dry biomass per hectare, but they're hoping they can increase that yield through breeding.

"Our goal is to get it consistently up above 20 tons per hectare, and if we can do that, then sorghum becomes an economically feasible biofuel," Kantor said.

This would be a boon not only to our effort to reduce fossil fuels—the sorghum project, specifically, got funding from the Department of Energy—but also to areas of the US where soil isn't rich enough to grow other crops. The beauty of sorghum is that it can grow just about anywhere.

But FarmView has multiple projects that adopt technology developed in other spheres and reverse engineers it to benefit farmers. Kantor told me about another effort that uses a computer vision system to look at a grape vine and count the number of grapes and leaves.

A vine without enough leaves means the quality of the grapes is low, but too many leaves mean the grapes don't get enough water, so that ratio is essential for grape farmers and determines when and how much they water their crops. By using this vision system while driving through the fields, a farmer can easily get all that data and make more prudent water decisions.

While investigating emerging technology applications on the farm is hardly a new idea, the cooperative FarmView program is a relatively new partnership, bringing together all kinds of researchers to try to solve our most pressing food security issues.

"This idea of taking the technology out into the field to solve real world problems is the thing that really drives all of us here," Kantor told me. "It's what really drives me."

Subscribe to pluspluspodcast , Motherboard's new show about the people and machines that are building our future.