For many, considering the effects artificial intelligence (AI) may soon have on society is a source of both anxiety and wonder. Agriculture, as much as any industry, is in line for big changes. Farm equipment may soon have a mind of its own.

The term AI, as it relates to agriculture, is often lumped in with other emergent technologies like autonomous equipment and field sensors. But, AI-based equipment is distinct in that rather than being programed to perform a function, it’s being designed to interpret data pulled from the field, act on it and teach itself best practices in the process.

The reason the terms are conflated is because AI will likely provide the foundation on which truly autonomous equipment is built and fields sensors will be its eyes and ears. (For more on the future of autonomy in ag, click here)

The first step to incorporating a new technology is understanding it. To this end, Farm Equipment asked manufacturers, educators and engineers how AI is reshaping the agricultural industry. Where is the technology today? Where is it headed? What obstacles lie ahead?

Why Is AI Important?

Just because you can do something doesn’t always mean you should. AI is meeting this argument in many of the industries it’s beginning to penetrate. Kraig Schulz, president of Autonomous Tractor Corp., thinks the agricultural industry will embrace AI and the resulting autonomous equipment because farmers’ margins are becoming razor-thin. He claims that if the average price of a bushel of corn is stacked against the average cost of producing it since 1980, the average income is negative $0.01 per bushel.

“Costs are up 60%, prices are only up 40% on average over that time period,” he says. “If you talk to the pundits, most don’t think this situation is going to get a lot better in the foreseeable future. Our view is that we have to keep trying to cut costs.”

Labor and equipment account for about 25% of farmers’ expenses, he says. Smarter, more efficient equipment can stretch the dollar in both categories.

If AI-based equipment results in smarter farming, it can cut costs in inputs as well. Scott Shearer, an Ohio State University ag engineering professor, says this is already being done with herbicide application. He points to the example of Blue River Technology, recently acquired by John Deere, and its development of machine learning in agricultural spraying equipment.

“They’re recognizing weeds and only treating the weeds in a field rather than a blanket application,” says Shearer. “This is a technology that really could change and reshape plant genetics and how some companies are focused. Tests have also shown about a 95% reduction in herbicide usage and being able to still control the weeds using only 5% of chemicals that we’ve traditionally used.”

Where Is AI Today?

It’s a misconception to say that AI is coming to farm fields because, to some extent, it’s already there. Alex Purdy, the head of John Deere Labs, based in San Francisco, says the equipment manufacturer has already been using machine learning to make real-time decisions in the field for several years. Purdy points to Deere’s interactive combine adjustment capabilities as an example.

“The Interactive Combine Adjust product we have today takes information from cameras that are embedded in the combine that sense things like grain damage and quality, straw condition, engine settings and operational characteristics,” he says. “Then, it runs an algorithm to make recommendations to the operator to change certain settings to maximize the growers’ desired outcomes whether that’s to increase fan speed or make another important combine adjustment. That process is already impactful today and affects the final yield. You get a lot less waste in the grain tank and you get less lost corn. That’s an example of this technology already at work.”

In short order, Purdy expects to see added computing power and more sensors on combines lead to further improvements such as feeding the farmer setting recommendations that significantly improve grain quality and machine productivity during harvest. He also speculates that the pace of advancement in agricultural AI is about to quicken as driverless car technology paves the way.

“Graphic processing units and sensors are becoming much more affordable, and that trend will continue being led by the driverless car segment,” says Purdy. “With the increased availability, decreased cost and growing strength of the related technologies, I think AI and machine learning will be incredibly transformative in agriculture going forward.”

Also championing Blue River Technology, Purdy claims that their sprayer solution allows farmers to go beyond making decisions at a field and sub-field level all the way down to the individual plant level. He notes that although Deere is still “a ways” from being able to employ this at “scale across the entire production system,” cotton farmers are already using the technology with some success.

“Using the same kind of AI technology that phones do for facial recognition, Blue River is helping cotton farmers identify individual plants like identifying palmer amaranth apart from cotton plants,” he says. “Then, the machine can go and take an action based on that information. I think that’s a clear example, but only the tip of the iceberg on ways that we can enhance the farmer’s ability to execute the job very effectively in the field at a plant-by-plant level.”

In addition to the implications this has for input costs, Purdy says less experienced operators could still be ensured optimal results if they’re supported by a battery of AI-based equipment that has its own experiences.

“You can have someone who’s new or who has a hired laborer in the cab and you can have confidence that they’re going to be doing the right thing because they’re backed up by a system that has traveled hundreds of thousands of acres and has that experience behind it,” he says.

Purdy says the technology today is already on the cusp of unlocking new efficiencies from “seedbed preparation to harvest.” The real-time sensing equipment on the market already does a good job of determining field specifications such as required down force during planting, he says. But, with AI analyzing conditions and teaching itself the optimum responses, equipment will soon be able to offer the farmer highly informed suggestions.

“Today we set the specification once and then run the equipment at that,” Purdy says. “We think there’s an opportunity to adjust the specs continuously based on real-time conditions. There is significant opportunity for optimization there.”

What’s On the Horizon?

Industry leaders seem to agree that we’re still years away from a swarm of light-weight fully-autonomous planters and combines rolling across the field 24/7 with crop scouting and spraying drones buzzing overhead.

Shearer says the industry will likely first see “supervised autonomy” as AI creeps into agricultural equipment.

“These are going to be machines that have a human watching over them,” he says. “But, hopefully we get to a point where that human watches maybe 10-15 machines at once. I think that’s going to be the progression. As we learn more about how to give that machine in the field intelligence, we’ll reduce our reliance on the human monitor.”

Matt Rushing, AGCO’s vice president of the Global ATS product line, also believes the slide into AI-based agricultural equipment will be a slow progressive one. He says advances in sensors are already laying the groundwork and that eventually, data pulled from the field will be fed into algorithms and real-time analytics engines that will calculate the optimum response to field conditions and act automatically.

“In the future, we’re going to see additional expansions with sensor technologies and the algorithms supporting them so eventually a human doesn’t have to even see the raw data or recognize that something’s happening before an action is taken in the field,” he says. “If sensors show there’s better moisture or organic matter in an area of the field, the technology will determine that a planter should raise its seeding population and notify the operator that it has done so.

“Or maybe a sensor will detect a pest on a leaf, and instantly apply pesticide? The sensors will supply the information and through the technology on-board, the equipment will react with an appropriate course of action.”

Shearer believes as AI in farm equipment improves, it’ll likely help transform drones into the valuable crop scouts the industry has always hoped they’d become.

“I truly believe there’s going to be opportunities with AI to get closer to identifying what some of the crop health problems are with fly-overs,” he says. “That’s not going to be the only thing though. We still rely heavily on our crop scout to go into the field and walk it, looking in the crop canopy to make their assessment. How do we position machines in the future to do the same thing?

“A lot is going on with rapid infield phenotyping right now. People are looking at putting robots within the canopy if you want to think of it that way. But again, how can we do this in a practical manner and how can we be cost effective? Those are going to be the important things to consider in the process of adopting AI.”

As to where AI innovations will come from in the future, Shearer thinks that tech startups from Silicon Valley in the mold of Blue River Technology will help lead the charge. But, he suggests that it may take an understanding of large scale farming to drive adoption.

Adoption Hurdles

Naturally, some of the biggest hurdles AI-based equipment will have to overcome are technological ones. New solutions will need to be rigorously field-tested and polished until they’re mature enough for widespread commercial use. Purdy even suggests that the large amount of computing power AI solutions will require may necessitate more electrification on field implements.

“Electrification generally is going to be required for smarter equipment,” he says. “The Blue River Technology that is pulled behind a tractor today requires electrical power and we expect the electrification of implements will continue to be an important requirement and enabler of tomorrow’s smart machines.”

Perhaps an equally large hurdle is a psychological one. Rushing notes that farmers are skeptical and not likely to leap out of the cab at the first mention of a smart machine that claims to know their farm better than they do. However, provable return on investment has a way of changing minds.

“Farmers can be the biggest skeptics in the world, especially when it comes to new technology,” says Rushing. “So I think you’ve got to have something that’s demonstrable and you have to have provable facts to back it up. I think once you have that, though, and can prove the waste and yield benefits, you’ll see more and more farmers adopt it.”

How Will Business Change?

Farmers still have a crop to grow and dealerships still have equipment to sell, but what effects will AI have on their business relationship? In terms of selling the equipment, Rushing says, not much. Proving the value of a purchase to the farmer will remain the golden rule as long as there is equipment to sell.

“If you can show the farmer that there’s value there, and it’s got to be a value that’s not just focused on waste, but more on yield, because they have realized they can’t cost reduce their way to prosperity,” says Rushing. “They’re looking for ways that they’re going to be able to somehow increase yields. If you can demonstrate the value of the sensors and technologies, farmers will use them and realize the benefits.”

Rushing also believes AI-based equipment will likely add revenue to dealerships’ service departments.

“Most of the revenue in a dealership is generated through parts and service,” he says. “All these new capabilities realized on the machine will be coupled with additional value added services that you can include at the time of purchase. In the future, in addition to buying a new planter with all these valuable sensors and technology, they’re also purchasing a recurring service package that ensures the machine and technology is optimized and available when it is needed. There will be more reliance on equipment dealers and technology specialists to make sure everything is working the way it should.”

What AI-based equipment may mean for the industry long term is as unknown as the technology’s applications themselves. Although, Shearer speculates that rapid advances and shrinking equipment may mean a shorter shelf life.

“We’re probably going to go to a different service life for implements and tractors,” he says. “Many of the tractors being built today probably have a life of 20,000 hours. I would expect to see that reduced to something like 5,000 to 6,000 hours. This is where technical obsolescence will meet mechanical life. I think farmers in the future will purchase technology on agricultural field machinery to do specific jobs. But, newer technology may render the previous technologies obsolete.”

Forecasting further, Shearer envisions a future where less equipment ownership is necessary on the farmer’s end. That could come in the form of leasing equipment or contracting for the service of AI-enabled equipment, he says. Either way, he feels that it’s likely more manufacturers may end up marketing their equipment directly to the end user.

He says those changes could have a serious effect on how dealerships operate, but the need for specialized service isn’t going anywhere — even if it changes shape.

“It’s going to be essential as AI makes supervised autonomous equipment possible to have service available 24/7,” Shearer says. “Farmers aren’t going to settle for the tractor being down for two days because they couldn’t get someone out to service it. However, with smaller equipment you might be able to warehouse all the parts essentially in the bed of a pickup truck. That changes the dynamic a bit in terms of being able to service the needs of a large fleet of smaller equipment. There’s going to be some interesting opportunities in rural America for new businesses, and these are going to be technology-based.”

When Will Artificial Intelligence Become A Natural Ingredient In Ag Equipment?

Universal Data Exchange Platform Offers Broad Potential for Mixing & Matching Ag Systems

What is Driving the Driverless Momentum in the Ag Equipment Industry?

Is Diesel’s Half-Century Farm Reign Ebbing?

Your Dealership Could Soon Be Printing 3-D Parts

January 2018 Issue Contents