How self-driving tractors, AI, and precision agriculture will save us from the impending food crisis

This article was originally published as a TechRepublic cover story.

Marcus Hall was nine years old when he first drove a tractor on his family's sprawling Iowa farm, eschewing Tonka trucks and Matchbox cars for long rides on heavy machinery. Growing up on a multigenerational family farm is common in an agricultural state like Iowa, where nearly 27 million acres are devoted to cropland--out of the 35 million acres that make up the state. For many, the experience ties them to the earth. The sense of freedom in an open field entices each season.

Hall grew up with all the trappings of a future farmer, but a penchant for technology led him down a more experimental path--to the test farm of ag equipment giant John Deere. As manager of the test farm, Hall gets to run field trials of John Deere's high-tech farm equipment before it goes to market. It's the proverbial dream job for a self-described farm boy.

"I just enjoy being out on the tractor," says Hall. "Plus, it's fun being part of this type of technology and the leading edge of what's out there."

It's a warm, breezy day in late May 2018, when we meet up with Hall at John Deere's test facility in Bondurant, IA. The farm sits on an unassuming patch of land framed by two-lane roads. Blue skies and a quintessential Midwest flatness consume the horizon. Tractor engines form a steady hum in the background.

This is where John Deere runs operational tests of its next-generation precision farming systems--massive tractors and combines that plant and harvest crops in thousand-acre fields.

Hall slides into the driver's seat of a new John Deere row-crop tractor, preparing for our test drive. His seat is flanked on the right with an array of iPads, touchscreens, and control panels, which slightly obscure a wraparound windshield encasing the two-seater cockpit. With a few taps on an iPad, he checks that the system is ready for seed disbursement, citing terms like row population, back pressure, and singulation. The tractor eases into motion, and after a few more control checks, Hall sits back, hands folded comfortably in his lap.

The tractor is driving itself.

A new era of farming

By most accounts, John Deere's steel plow is one of the earliest examples of farm technology built in modern times. Many years and a technological boom later, the farm has grown into a launchpad of innovation, with some of the first high-impact use cases of GPS guidance and self-driving vehicle systems taking shape among the crops.

At the core of this new era in farming is precision agriculture, a farming management concept that uses technology to observe, measure, and respond to field variability in crops. As the concept has evolved, it's allowed farmers to view their cropland as sub zones within a field to optimize equipment and supplies like fertilizer, herbicides, and water.

More refined applications are trying to take that precision all the way down to the plant level, allowing farmers to understand and treat plants individually.

In John Deere's case, innovation has ramped up since its purchase of artificial intelligence (AI) startup Blue River Technology in 2017. The acquisition cemented John Deere's arrival in Silicon Valley and now supports its efforts to incorporate machine learning, deep learning, and robotics into the brains of its precision farm equipment.

The goal is to use automated driving technology, computer vision, telematics, and cloud-based mobile applications to help farmers double or triple their yields--a feat that will be key to keeping up with global food demands as the Earth's population grows over the next 30 years.

"By 2050, there's going to be nine billion people on the planet," said Terry Pickett, manager of advanced engineering for John Deere's Intelligent Solutions Group. "And to feed those people, there's different estimates out there, but we probably need to increase current production by 70%. That's not very long away when you look at the number of years it takes to develop the type of equipment and technology we need. It's a real race."

The path to precision ag

In the late 1990s, John Deere became one of the first farm equipment vendors to bring GPS guidance to agriculture. Nowadays, GPS guidance is almost ubiquitous in large production ag and farming.

GPS guidance is arguably the most significant technological advancement that has come out of the precision ag movement over the last two decades, says John Stone, senior vice president for John Deere's IntelligentSolutions Group. The reason GPS is so important, Stone posits, is because it's what makes both precision and automation possible.

Driving a tractor isn't like driving a car--there are no yellow and white lines to keep drivers in check. To plow straight, the farmers of yore used the hood ornament--akin to a gunsight--to line up the tractor with a distant landmark, like a tall tree or a jut in the hillside. By aiming for that landmark, the farmer could keep the tractor's path straight within a reasonable margin of error.

But a tractor is usually pulling a plow, or some other implement, upwards of 30 feet wide, meaning the driver not only had to stay focused on the distant tree, but constantly turn and look to make sure the plow was still in line.

SEE:The future of food (ZDNet/TechRepublic special feature)

"If I'm a tractor operator, I need to be more concerned with what's going on behind me," said Stone. "Planting seeds in the right place, and doing that job with the right precision, is far more important than, 'Am I steering the tractor in a straight line?' So we let the technology, the GPS guidance, do that for you. And the operator can focus on more high-value added tasks."

As for the precision aspect, GPS receivers built by John Deere provide navigation accuracy down to one inch. In the context of farming, that accuracy is critical to making sure every seed is in the right place, with the right depth, soil contact, and spacing that it needs to grow into a food-producing crop. The combination of precision and automation has already made a consequential impact on the job of farming.

"Today, precision ag touches on every little bit of our operation," said Jamie Blythe, owner of the Blythe Cotton Company in Town Creek, AL. She grew up fifth generation on the farm and became a formal partner in operations when she was 18. "I was probably one of the last generations to actually chop cotton in my area, and since I was a kid things have changed drastically," she said.

SEE: Glossary: Smart farming

Blythe and her husband work 3,500 acres on what she describes as a midsize farm for her area. In 2007 they implemented John Deere's AutoTrac guidance system on their equipment, and now utilize an array of technology geared toward equipment and input, monitoring, and data collection and analysis.

With these systems, Blythe said crop yields have doubled over the past 10 years. Some of the gains are due to to better agronomics, seed selection, and use of hybrid seeds, but they're also attributable to land optimization thanks to precision ag.

"It's no longer possible to manage a field as one complete entity," Blythe said. "Due to the variability in our soil types, elevation, and microenvironments, we now manage fields in zones that allow us to tailor inputs to the specific yield potential of those parts of the field. Precision ag technology has made a huge difference in helping us make a quicker transition to the more modern style of management."

IoT and AI take root

The Internet of Things (IoT) is tangible for today's farmers. The machines farmers employ to traverse their fields are stuffed with sensors and software that gather data, process it with machine learning, and beam it into mobile apps. The sensors are the eyes of the machine. The software and mobile apps bring the data to life.

On John Deere's MyOperations app, for instance, a machine operator can see the five most critical settings of a combine while it's harvesting a field of corn. If a setting is out of whack, the operator can adjust the settings of one or multiple combines remotely from a mobile device.

"In many cases, farmers are operating across 10 to 15 mile radiuses for their fields," said Lane Arthur, director of digital solutions for John Deere. "Obviously they can't be in every spot at once. They need the technology and connectivity in order to see in real time what's actually happening. That's critical for the on-the-go decision making that farmers face."

John Deere began using GPS regularly around 1997. Roughly 20 years on, IoT now has a firm foundation in precision agriculture. The next mission for ag-tech vendors is to apply AI to the three main steps of the farming process--planting, spraying, and harvesting. Each of these steps has a material impact on a farmer's output and productivity.

The planting stage is especially intricate, sort of like performing surgery on dirt. A tractor with a planter attached creates beds where the seeds are planted; and furrows, which are the narrow trenches between the beds. As this happens, seeds are shot out at a certain cadence to ensure proper depth and spacing. If a seed fails to land in the right spot, it doesn't get watered and fertilized efficiently, and most likely doesn't grow.

With AI, John Deere envisions a planter that can be made to understand the ground conditions and alter planting settings automatically.

Must read: Download this article as a PDF (free registration required).

The spraying step needs to make the same type of tiny precisions as the planting step. Weeds cost farmers an estimated $11 billion a year, according to the US Environmental Protection Agency (EPA). For John Deere, this is where the motives for its acquisition of Blue River become obvious. Blue River made the See & Spray, a weed control cotton machine that's capable of seeing the ground and distinguishing between crops and the weeds, down to a level of millimeters. It then zaps the weeds--and only the weeds--with herbicides to kill them off.

"Precision spraying solves a bunch of herbicide resistance problems for cotton growers," said Willy Pell, director of new technology for Blue River. "It saves them about 90% on herbicides; so it's better for the environment, better for their pocketbook, and gets a better result as well."

Pell said the harvesting step is due for a similar AI treatment. Eventually, harvesters will have the brains to adjust their blades to a custom setting for individual stalks, rather than operating on a single setting for an entire field.

"The really, really exciting part is closing the whole loop," Pell said. "Think of A/B testing for agriculture. We can induce variation at various parts of the grow stage, and run experiments all the time to generate agronomic knowledge, so that every year Deere equipment gets put out in the field, it does it better and better and better."

SEE: The future of food (ZDNet/TechRepublic special feature)

"The next stage of precision ag is to go down to the plant level," Pell continued. "With cameras, with onboard computing, computer vision, and machine learning, we can understand what that plant needs and give it exactly what it needs right then and there. The potential upsides to this are yield doubling with the same machine."

A farmer's uncertainties

The farming advances John Deere and others in the precision ag industry are working on tie into a range of issues facing today's farmers. Some of the ag tech benefits are blatant, like saving farmers time and money, and boosting average crop yields; but others stem from more complex global concerns, like weather uncertainties and labor shortages, and a rising demand for healthier, sustainably grown food.

Adapting to weather changes is a constant in the job of farming, as severe weather events brought on by rising temperatures and atmospheric changes are known to cripple crop yields. Recent analysis from the National Academy of Sciences that looked into the climate impact on crop yields produced sobering results. The study's authors argue that for each 1° Celsius rise in global mean temperature there would be a 7.4% decrease in yields of corn, a 6% decrease in yields of wheat, and a 3% yield decrease in rice.

Modern farmers now look to technology and data to understand and mitigate the impacts brought on by climate variability. Even with a perfectly balanced scenario of seeding, spraying, and harvesting, the inevitable unknowns--be it floods, droughts, or pests--always remain a threat. But agronomic data gathered by precision farming systems allows farmers to stay ahead of these threats and work around problems when they arise.

Must read: Download this article as a PDF (free registration required).

"We see data really helping farmers make more informed decisions around what they should be doing," said Arthur. "In the event of bad weather, for example, my hope is that our sensors and our technology will allow them to adjust and adapt. Even in a bad scenario, precision ag really helps the farmer preserve and conserve the value they have in their field."

And then there's the labor dilemma. From 2005 to 2019, an estimated 58 million fewer people will be employed in agriculture, a decrease of 11% of the agricultural workforce, according to data from the Global Harvest Initiative. Similarly, more humans are moving to live in cities, compounding the workforce shortage on farms.

Meanwhile, the average US farm clocks in at 434 acres, or roughly 330 football fields, per the US Department of Agriculture (USDA). Boiled down, this means more acres of cropland are at risk of being underutilized or abandoned due to a lack of farm labor.

Back at the Blythe Cotton Company, 3,500 acres are maintained by a total of five people--the husband and wife ownership team and three laborers. Jamie Blythe recalls a time of cheap, plentiful labor when she was growing up on the farm. But then diesel fuel became more expensive, and people started leaving the area. Blythe said the farm was forced to evolve, adapting a more streamlined operations approach to get work done, and, in the end, survive.

SEE: The future of food (ZDNet/TechRepublic special feature)

"My husband and I operate the equipment with three other guys," Blythe said. "As a result, we need to maximize the use and efficiency of our equipment and technology. This has been a very gradual, organic evolution, but at some point I looked around and realized this is the only way we're going to stay in business."

What's even more important than yield increase is the increase of profit by acre, explained Blythe.

"The net profit per acre is extremely important because the amount we put in per an acre to increase yield and to make sure it's appropriate is what's going to keep us in business," she said. "Equally important are time and energy saved, not just in diesel fuel but also in physical energy. There are very few of us out on the field, and we need to be going anywhere from 12 to 20 hours a day and still come home to our families. If we can maximize the amount of energy we put out and still have energy to play with our kids at the end of the day, then that's the biggest benefit for us."

Food safety, nutrition, and sustainability are also driving forces behind innovations in ag. Data from the Global Harvest Initiative shows that a rising number of consumers want to ensure their food is sustainably produced and many are concerned about the safety, price, and availability of nutritious food.

"Diets are improving, and that's compounding the need and demand for grains," said John Deere's John Stone. "We've got to find a way to meet that need. Precision technology, we feel, is the way to do that."

Precision agriculture can also help the food industry to become more sustainable. For instance, data generated by sensors allows farmers to see areas where soil moisture has dropped and prevents them from wasting water in areas that don't need it. Sensors can also collect data about soil nutrient profiles, enabling farmers to use precise fertilizer applications only where the soil requires it.

"We use water sensors to measure soil moisture and various depths in the soil profile," said Blythe. "Less than 10% of our acres are irrigated, but those sensors help me make decisions to utilize our water resources in a more sustainable manner."

Problem solving with AI

A farmer often says they have 40 shots to get it right in their lifetime. They start farming around age 20 and aim to retire around age 60. Every year is a new experiment, with unseen variables they can't control. Maybe the ground's a little different, or the weather patterns change, or they're trying out different seeds and new equipment.

The role of AI in all of this is to help farmers learn from the sort of A/B tests that are happening all around them. An AI system can look across a farmer's operations to analyze data on weather, temperature, moisture, and soil composition, and provide insights on how to optimize equipment, improve planning, minimize waste and increase yields.

"We have the sensors in all of these parts of the farming job, and now we can actually use AI to begin to have insights about what is the best way to approach farming challenges given certain conditions," said Julian Sanchez, director of the John Deere European Technology Innovation Center. "It's really impossible to do without something like AI because otherwise, you're just deterministically trying to understand multivariate equations, endlessly."

Consider the need to boost crop yields. The Global Harvest Initiative predicts that the world's population of 7.3 billion will grow to nearly 10 billion by 2050. Some estimates, such as the one cited by John Deere's Terry Pickett, puts that figure closer to 9 billion. In either case, there's consensus that to feed those additional people by 2050, the global food supply needs to at least double, and global agricultural productivity needs to increase by 1.75% a year.

Must read: Download this article as a PDF (free registration required).

How close we've come to meeting these food production goals depends on the statistic. One barometer raising concern is what's known as total factor productivity (TFP), a measure of agricultural productivity that takes into account all of the land, labor, capital, and material resources utilized in farm production and compares them with the total amount of crop and livestock output. The 2017 GAP Index reveals that for the fourth straight year, global TFP growth is not accelerating fast enough to sustainably double agricultural output by 2050.

When put into a historical context, however, the overall productivity gains are promising. According to the US Department of Agriculture (USDA), in 1940 the average American farmer fed 19 people a year; today the USDA estimates that the average American farmer feeds 155 people. Breaking it down to individual crops, corn yields in the US grew by 61% between 1980 and 2015, while soybean yields improved by 29% over the same period.

SEE: The future of food (ZDNet/TechRepublic special feature)

It's a similar story with labor. In the early 1800s, it would take about 300 hours of labor to produce 100 bushels of wheat. Today, with the application of technology and AI, it takes less than an hour to produce 100 bushels of wheat.

Nevertheless, there's a consensus that farmers will be challenged to do even more in the years ahead, and technology will be key to making it happen.

The connectivity challenge

The precision ag industry is aiming high when it comes to food production targets, but the technologies being developed right now cannot exist effectively without the proper infrastructure. For high-tech farming systems, the infrastructure challenge boils down to connectivity, and in many rural settings, the lack thereof.

In rural America, 5G technology is a promising connectivity development. The fifth-generation wireless broadband technology is expected to bring low-latency wireless speeds of up to 1GB/s to parts of the country that typically lack coverage.

While 5G is not yet widely available, John Deere is optimistic about its potential to support new precision agriculture capabilities on farm equipment with real-time connectivity. High-speed connectivity is essential to the technology inside John Deere's farm equipment. John Deere says that every new machine that leaves its factory has a 4G LTE modem, Wi-Fi, and Bluetooth.

The idea is to have farm equipment that's able to communicate with other machines on the field by streaming data from vehicle to cloud and back down to machine operators in the shortest time possible. Machine-to-machine communication is what helps farmers prevent row overlap if they're running multiple machines at the same time. It also lets operators share data with each other for things like machine optimization.

With connectivity the machines become socially aware, knowing exactly where other machines are located in relation to themselves.

"So you might have two combines in the same field harvesting the same crop," said Sanchez. "Right now we have to assume there will be 30 to 60 second delays in communications between these combines while they upload data to the cloud, have it processed, and downloaded to the other combine so they can share information."

With 5G, Sanchez said communication between those two combines through the cloud could be cut down to less than one second.

Must read: Download this article as a PDF (free registration required).

Of course, there are vast infrastructure challenges to reaching this kind of connectivity utopia in America's heartland. Telecommunications providers caution that 5G wireless could struggle in rural areas, particularly ones with lots of trees and foliage, and could encounter other issues due to low population density. As a result, adoption of precision agriculture systems is limited in many rural areas due to a lack of broadband service.

Although we're still years from realizing the true potential of rural 5G, the precision ag industry plans to continue making advancements to farm technology until US telcos and regulators catch up.

The road ahead

Food security is an issue facing all of humanity. The bleakest predictions suggest that the world is careening toward a dystopian future where food shortages prompt riots and wars. It's an unlikely outcome, but still motivating for companies like John Deere. The farm equipment giant is rallying around emerging technologies that will improve its precision farming systems and help stabilize the world's food supply for the years ahead.

Executives from John Deere are now known to roam the floors of CES, the massive technology conference held each year in Las Vegas, looking for new moon shot ideas to bring into the fold. In the near term, John Deere is focused on ways to incorporate computer vision, augmented with deep learning algorithms and AI, into next-gen precision farming systems.

"We also monitor quite a bit the progress in automotive, especially things around the connected vehicle," said Sanchez. "If you track our capabilities in having mobile devices as part of our vehicle ecosystems, it drags very, very closely with the connected vehicle, the connected model."

Also: The future of food (ZDNet/TechRepublic special feature)

John Deere is also watching developments around microelectromechanical systems (MEMS), nanoelectromechanical systems, and nanotechnology. These technologies tie into the sensors used in precision ag to collect different types of information.

"They give us capabilities that are unimaginable," said Blue River's Pell. "The question is, which ones will become popular, and can use them in a low cost manner? We try to focus on one technology and watch it swirl, and then we'll pick it up at some point when it's very valuable to us and our customers."

But technology advances in the private sector only represent one side of the global food supply equation. In the US, some of the responsibility falls to the Department of Agriculture and funding it allocates via farm bills toward research and development of sustainable production technologies.

SEE: Glossary: Smart farming

Experts within the SoAR Foundation, which advocates for agricultural research funding, say the USDA's annual research budget needs to increase substantially from today's levels in order to revitalize the productivity growth rates of American agriculture and ensure the sustainability of the sector.

Regulators should also be conscious of land shrinkage and the impacts it could have on US food production. The American Farmland Trust projects that more than 40 acres of US farm and ranch land are lost every hour to urban sprawl or development. On a global scale, scientists also estimate a more than 2% loss of highly productive cropland due to urban expansion.

---

Must read

• The quiet tech revolution overtaking the dairy farming industry (TechRepublic)
• How drones are changing farming in rural Japan (TechRepublic)
• More must-see food tech coverage (TechRepublic Flipboard magazine)

---

Technology advancements also present an obstacle for farmers. The team back at the Blythe Cotton Company admits to the learning curve they still face when adopting new technologies. In the end, it's up to the men and women out in the field to make sense of precision ag technology and how to truly utilize the systems to their potential.

"The scope of precision ag is so tremendous, and there's so many different applications to it that I feel like we're still getting our toes wet," said Blythe. "In 20 years from now, precision ag will drastically change the way we farm."

Must read: Download this article as a PDF (free registration required).

11 delivery robots that will soon carry food and packages to your door

Related stories:

• Smart farming: How IoT, robotics, and AI are tackling one of the biggest problems of the century
• Agriculture 4.0: How digital farming is revolutionizing the future of food
• Meal replacements make tech workers more productive. Can they also solve world hunger?
• The world's most important cold storage facility safeguards the future of food
• Top 5 ways technology is changing food