SC researchers explore AI robots for farming

Farmers in South Carolina may receive extra help in the future from an unexpected source: AI-fueled robotics.

At least, that’s the hope of Joe Mari Maja, the senior researcher and director of the Center of Applied Artificial Intelligence for Sustainable Agriculture at S.C. State University.

“Farming in the future might be farming without farmers,” Maja said. “Farmers will probably be staying in their house, facing the computer and telling the robot, ‘Hey, do this thing now.’”

His mission aligns with the United Nations’ Sustainable Development Goals, using AI to improve agricultural practices, sustainability and equity.

That vision is already taking shape in Olar, South Carolina, where Maja and his team apply AI to real-world farming problems, such as using drones to track peach tree defoliation and building autonomous cotton harvesters. 

Maja knew that AI, when it became popular, would be a major asset in his research on precision agriculture, especially when it comes to crunching numbers and modeling.

“You input everything into AI, and then it can create its own model. You don’t have to do anything,” Maja said. “It gives you the results, and that’s just so impressive. It took you a year, but now you can do it in less than a month.” 

Maja’s aid to farmers reaches outside the farm. He hosts workshops teaching farmers how to build apps using AI. These apps can be tailored to a farmer’s needs.

He has paved the way for South Carolina farmers to access more AI-powered technology.

Still, the state isn’t leading the race by any means.

“We’re not falling behind,” Maja said. “But we’re not the front-runners either. For one reason, we don’t have the big infrastructure that AI needs. But we do our part based on what we have.”

But it’s not a competition, said Ed Barnes, the senior director of agricultural & environmental research at Cotton Incorporated, a not-for-profit research and promotion organization for cotton developers.

“Our goal is to make our farmers as profitable as possible, but we’re not really in a competition,” Barnes said. “What we are trying to encourage is collaboration.”

Teaching something to a machine via AI requires tens of thousands of images to be truly effective, such as for cotton weed identification. That’s why the projects and research that Cotton Inc. funds are encouraged to be open source, so researchers from different organizations can share their images and benefit from each other.

Cotton Inc. is based in North Carolina and has offices that cover China, Latin America and Southeast Asia. 

Barnes oversees a cotton harvesting project that Maja is leading for Cotton Inc. It’s an autonomous robot that uses machine vision to detect harvestable cotton on a crop. Cotton doesn’t mature all at once, but instead from the bottom to the top of the plant, which is why it would save time for a harvester to pick it as soon as it’s ready. 

South Carolina farmers would especially benefit from this project, Barnes said.

S.C. farms, on average, are much smaller than others in the country, for example. Standard equipment from large companies, such as John Deere, can be too large.

“There are smaller fields where it’s hard to fit the equipment into them,” Barnes said. 

The smaller robot cotton harvester that Maja is working on would be much more efficient. 

“If your big harvester dies, you’re not harvesting cotton,” he said. “But if we had, let’s say, three or four little harvesters, again, we’re less at risk from equipment failure. So, I envision one future where we see agricultural equipment getting smaller and autonomous.”

Cotton Inc. only recently entered the AI field. Maja was the first person it paid specifically to research AI in agriculture.

After recent successes, the company wants to expand its efforts, Barnes said.

“I’m an engineer, and engineers like technology way too much,” he said. “We might find a more complicated solution than necessary, but what I’m seeing is my other colleagues … making more use of AI-related tools in the last five years.”

There are still some concerns about the use of AI.

Farming practices can be improved through AI technology, but it’s not a perfect system, said Michael Cosh, a research physical scientist with the U.S. Department of Agriculture’s Agricultural Research Service.

“The challenge will be to use it as a tool to make better decisions while still maintaining control,” Cosh wrote in an email. “AI is only as good as the training dataset. So, it is up to the person using the tool to determine if the situation has gone outside of normal conditions, which is what AI is often trained on.”

In the short term, Barnes thinks it can be risky to rely solely on the technology. Farmers’ confidence in it might be low at first. 

“There are some concerns about AI, especially as we use it for data interpretation — making sure we know how the algorithms are coming to certain conclusions,” he said. “And I do worry a little bit about companies that will try to overhype their use of AI and what it can do for a farmer in the short term.”

AI projects aren’t cheap, either. Cotton Inc. has given nearly $1 million in funding to Maja over the six years they have collaborated.

Barnes said it’s worth every penny.

“Some of these things are now running on … small computers,” he said. “Computer power has gotten so great. Another factor that will help us on our sustainability goals … (is that) they’re based on electric motors. So, that enables better use of solar power.”  

The market for AI in agriculture is expected to more than double in value in the United States in the next five years, from $351 million to $705 million, according to a report from Research and Markets, which describes itself as the world’s largest market research store. 

It’s not yet commonplace to see AI used on smaller farms, and Maja acknowledges that.

“I know that there are a lot of people who are using artificial intelligence, but I just can’t pinpoint exactly where,” Maja said.

Nevertheless, Maja said he knows AI in agriculture is the future.

“I hope we start it here in South Carolina,” he said. “That’s my dream.”