Tech + Engineering


Farms of the Future Will Run on Robots and Drones

On a sunny day in an otherwise rain-soaked May, Forrest Watson, dirt caked on his work boots, kneels in the middle of his uncle’s cornfield and points at one of the thousands of knee-high stalks.

“This one,” he says, “should be over here,” poking the soil a few inches away.

It’s peculiar to think that Watson, 22, would be so particular about the location of a single stalk of corn among the 1,455 acres planted on his uncle Jeff’s farm in Avon, New York. But Mulligan Farm is a particular place. The tractors, which drive themselves, don’t stray from their paths by more than an inch. The planter, towed behind a tractor, knows the nutritional content of every square foot of every field. It plants more seeds in richer soil and fewer in the thinner stuff. Jeff Mulligan, the farm’s owner, hopes they’ll soon have access to small drones that can fly over the fields and monitor plant health from above.

self-guided tractor
John Huenemoerder, an employee on Mulligan Farm, demonstrates a self-guided tractor. The black cylinder around the steering column controls the steering, but disengages as soon as a human operator takes the wheel.

This year, Mulligan Farm crop manager John Huenemoerder took only one lunch break in three weeks of planting this season, but it wasn’t because he was skipping meals. Instead of stopping the tractor while he ate, Huenemoerder simply tilted the steering wheel up out of his way and ate his lunch in the cab, keeping an eye on the equipment from the cab. The tractor did the rest.

Welcome to the future of farming.

Decades of Changes

Mulligan Farm is just a few miles off Interstate 390, a flat rural highway that wanders south from the edge of Lake Ontario toward Pennsylvania—closer in every way to Ohio than New York City. The village of Avon hasn’t grown much since Mulligan’s childhood, when he used to ride his pony across the two miles of farmland between his family’s farm and Tom Wahl’s burger joint at the edge of the town.

Mulligan, now 57, took over the farm from his father in 1981 and decided to dedicate it to dairy production. He now has 1,200 cows supported by the more than two square miles of corn he’s planted this year.

When Mulligan’s grandfather, Edward Mulligan, bought the farm in 1920, it had 640 acres of fields tended by eight teams of horses. In the decades since, tractors and their myriad trailer attachments have made that work easier, but new machines can only do so much. As Jeff expanded, he began using all the fields available to him but found his yields still wanting, so he called his long-time consulting firm. They encouraged him to add more computational power and automation to squeeze the best possible harvest out of each acre.

Mulligan’s investment in technology has paid off. “We’ve seen better yields over the last two years than we’ve ever seen,” he says.

Echoes of Silicon Valley

Mulligan Farm is one of many farms leveraging high-tech equipment and precise field data to increase crop yields and efficiency. While the self-driving tractors make for a fantastic show, they are just the beginning. Precision agriculture is still in its early stages. If these were the early days of the personal computer revolution, Mulligan Farm would be a small garage in Silicon Valley in the 1970s. And like that moment in history, the possibilities for precision agriculture today are seemingly endless.

These new tools, though promising, aren’t ready for widespread adoption. Most farms—faced with wide-ranging, expensive, and constantly changing arrays of options—have been slow to buy in or unable to take full advantage. Mulligan’s tractors may drive themselves, but he still has to pay operators to babysit them. Some experts say the end-game for precision agriculture will be unrecognizable by today’s standards.

rmax drone
An operator guides a remotely-operated UAV at a vineyard in California. While farmers in the United States are now testing UAVs on farms, Japanese farmers have been using them for years.

Within the next decade, the massive specialty tractors and attachments that have made large-scale farming possible could be replaced by multiple small, driverless machines, says Carl Dillon, a professor of farm management at the University of Kentucky. Each of them will work on a single row of crops at a time, their paths varying by less than an inch. And they won’t just be crawling around the ground, either. Small drones will hover from plant to plant, dropping just enough fertilizer or spraying exactly the right amount of pesticide. In some places, that future is nearer than others. Drones are already used in Japan to treat small areas that are impractical or impossible to reach in a large, fuel-guzzling tractor.

Ever More Automation

Despite the drastic changes, the near future of American farming may, in some ways, more closely resemble the distant past. Instead of a massive machines slowly combing over vast swaths of land, scores of individual laborers will work their own small sections, one row, one plant at a time. The only difference is they will be robots, working day or night, continuously streaming data about growth rates, soil fertility, water usage, and more to the farm office.

With drivers out of the equation, robots will grow smaller and more nimble. “You can imagine a future where you’d have huge numbers of small robots swarming over a field,” says Martin Ford, an entrepreneur and author of The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future.

This level of automation and data collection hasn’t come to Mulligan Farm quite yet. Mulligan knows his annual yield, but he can’t say for sure how many bushels came from which acres. (He hasn’t invested in GPS-based yield monitoring, though other farms have.) Still, even if the data isn’t there, Mulligan knows the methods are working. Overall yields are up, even if he doesn’t know exactly where on the farm they are highest.

His farm’s adoption of precision agriculture was gradual. It began six years ago when they bought a GPS-based system, similar to the navigation units found in cars but with a small attachment mounted to the tractor’s steering wheel that keeps the machine on course. The operator can also steer manually—a bar of lights along the top of the screen tells the driver which way to adjust the wheel to stay on track.

This sort of incremental technology has been well-received in the agricultural community, where margins are traditionally so tight that tractors which stray from their course by just six inches can noticeably cut into profits. One study that Dillon worked on examined the potential benefits of adopting auto-steer technologies and found that the reduction in overlap, longer working hours, and faster passes could raise net revenue by 2.3%. That may not sound like much, but in agriculture—“where even an economic profit of zero is delightful,” Dillon says—it’s a large advantage.

More recently, Mulligan upgraded his guidance system. Early in the season, the tractor’s path through the field is programmed onto a USB drive. All passes follow that exact route with sub-inch accuracy, which is made possible by an advanced GPS receiver that taps into a land-based antenna system known as real-time kinematics (RTK). By ensuring such precision, corn seeds, for example, are placed exactly above the nitrogen fertilizer, which has been injected nine inches below the surface by the tiller.

night combine
Automated machinery could allow farmers to run their equipment day and night without the need for in-cab human supervision.

Alongside that system is another computer attached to the seed planter. In it is stored an electro-conductivity map of the field, which was generated by a specialized machine that drags weakly electrified metal disks through the soil to measure its conductivity. Since a soil’s electro-conductivity is related to its fertility, the computer tells the planter how frequently it should lay seeds. In the field’s less fertile areas, it will plant only 29,000 seeds per acre, but in other places it plants up to 40,000. Placing fewer plants in weak soil makes for fewer duds, while the lost density is made up for in the richer, more thickly-populated areas. “We may put the same amount on the whole field,” Mulligan says, comparing this year’s planting to years past, “but it’s going in the right place.”

Once the seeds sprout, Mulligan uses his fields’ reference data to tend his crops during the growing season. But there are some farmers who now continuously monitor plant health throughout the season. They fly small drones equipped with sensors to measure the amount of red light and near-infrared heat that is reflected by the leaves. The two reflectance values are fed into a calculation known as the normalized difference vegetation index, or NDVI; healthier crops have higher NDVI values. Such sensors can also be mounted on a fertilizer sprayer’s boom to ensure each sprayer delivers just the right amount of fertilizer depending on the health of an individual plant or section. One advanced model uses multiple sensors to make dozens of on-the-fly calculations per second to adjust the rate of fertilizer application appropriately—all while rumbling along at 20 miles an hour.

Farm-Brew Computer Club

Such highly-polished systems make it easier for farmers to adopt precision technology, but they are also costly. For those willing to tinker, though, there’s an entire maker movement within agriculture that’s hacking hardware so farmers get the most from their fields. Again, it’s reminiscent of the early days of Silicon Valley. In fact, the comparison is more apt than you might suspect, as some farmers are hacking small, customizable computers known as Arduino boards to monitor sensors, send alerts to their cells phones, and link together all the disparate pieces of a precision ag operation.

Somewhere in between the sleek-but-expensive commercial products and the do-it-yourself hacks are a series of devices being developed by Louis Thiery, a Cambridge, Massachusetts-based entrepreneur at Apitronics, LLC. Thiery is creating an Arduino-compatible device tailor-made for agricultural and environmental uses. He was inspired after helping a farmer hack together a simple alert system: when something wasn’t right in the greenhouse, it would send the farmer a text-message. “Basically,” Thiery says, “he couldn’t find anything that was within the price range he wanted that did what he needed.”

Now, Thiery is working to create a flexible system geared toward small farms. Using a wireless hub that connects to the farmer’s modem—a “hive” in Apitronics’s lingo—the farmer can deploy various sensors and other electronics—they call them “bees”—around the farm. One Apitronics bee may monitor the greenhouse, while others keep track of soil moisture in the fields.

Like the open-source, hacker-oriented Arduino project on which Thiery’s platform is based, Apitronics doesn’t dictate how to use its products. “I’m trying to make something that’s versatile that will work for a guy that’s only working 100 to 500 acres,” Thiery says. He has a crowdfunding campaign set to launch in September that he hopes will lead to Apitronics’ first production run. After that, he wants to move beyond sensors and alerts to simple automation. Instead of merely sending an alert that the greenhouse is too hot, Thiery wants one of Apitronics’s bees to tap into the greenhouse’s climate control system and open the vents or turn on the irrigation, depending on the parameters set by the farmer.

A farmer assembles Fido, a do-it-yourself Arduino-based alert system developed by Louis Thiery.

Thiery likens the introduction of his open agricultural hardware to the early days of Linux, an open-source operating system beloved by hackers. People are trying to work out which parts of precision ag are most “tweakable,” he says.

Thiery is not the only one exploring the intersection of silicon and soil. Dorn Cox, who met Thiery through Farm Hack, a community of engineers and farmers focusing on these DIY projects, is outfitting small drones with modified point-and-shoot cameras to measure NDVI on his 250-acre family farm in southern New Hampshire. “For well, well under a thousand dollars, you’re set up with a high-resolution aerial mapping setup,” Cox says.

Low-cost solutions like these, as well as more expensive automated equipment like Mulligan’s, have environmental benefits, too, by making more food with less land, less fertilizer, and fewer pesticides. For example, by injecting nitrogen fertilizer into the soil nine inches below the surface and planting the seeds right on top, more fertilizer makes it to the plant and less runs off into rivers, lakes, and streams.

“How you farm [has] a profound effect on the environment,” says Rich Wildman, co-founder of Agrinetix, the farm consultancy used by the Mulligans. Indeed, a 2012 study from the University of Kentucky found that precision agriculture practices can reduce a farm’s greenhouse gas emissions by up to 2.4 percent. That is no small amount, considering one-third of all greenhouse gas emissions come from farming.

Bringing Automation to Everyone

With Mulligan Farm reporting their best harvests in years, why isn’t every tractor in America driving itself? “Because of the cost, the complexity, other issues, some farmers won’t adopt this technology,” says Dillon, the University of Kentucky professor. Mulligan and his employees agree, saying that some farms are barely scraping by and have little reserve capital to invest in new equipment.

Agricultural technology can have a steep learning curve, Dillon says, and it is only worth the cost if farmers see a return. Unfortunately, he says, “It doesn’t always make money.” Furthermore, when farms invest large sums of money on new equipment—Mulligan Farm has spent over $100,000 on precision agriculture upgrades over the last six years—they want to be sure they can use it to its fullest capacity. Given the dozens of new products introduced every year, evaluating their capabilities is no small task.

That’s where agricultural extension agents come in. Universities, government agencies, and firms like Wildman’s Agrinetix stay abreast of the newest equipment and methods so farmers don’t have to. Agrinetix, for example, works with about a dozen manufacturers of precision agriculture equipment, reselling their products to clients based on the particular needs of the farm.

For some, today’s precision technology is still too expensive to be worth the cost. Yet as smaller, more modular equipment becomes widespread—starting with Apitronics’s bees on up to drones that deliver fertilizer and pesticides—even the leanest operations may be able to partake. Small machines, Dillon says, can be a “scale-neutral” technology: In the past, agricultural advancements focused on making tractors bigger and implements larger, say a 10-bladed plow versus a 5-bladed model. For large farms, the 10-bladed model saved them time and fuel. But for smaller farms, the benefits weren’t always worth the costs; they just didn’t spend that much time plowing.

The proliferation of smaller machines could reverse that bias and democratize precision agriculture. “In fact, these machines will potentially—especially in the short run—benefit small- and medium-size farms more than they would on large farms,” Dillon says. “It’s more scalable to the size of the operation. You can spend closer to what you need.”

As costs fall and further advances make precision agriculture more lucrative, the pace of adoption will quicken. Today’s precision farmers have found ways to make the tools and techniques work for them, but much like the early days of desktop computing, the mainstream has yet to follow.

In 2020, Mulligan Farm will celebrate its 100th planting season. Mulligan may have entirely driverless tractors by then. They’ll be tilling fields and wirelessly transmitting location data to the planter, which will use an electro-conductivity map to plant the seeds just so. After the seedlings nose their leaves above the soil, perhaps a small drone will hover above the field measuring NDVI. That data will be sent to the sprayer, which will use it to judge how much fertilizer to apply. At the end of the year, the harvester will pass over those same rows, taking even more measurements as it harvests the corn.

By the time Forrest Watson, Jeff Mulligan’s nephew, is his uncle’s age, he could be working the fields of Mulligan Farm without ever getting his boots muddy. It’ll be a far cry from the eight-horse teams that worked the land nearly a century ago.