Drones, Joysticks, and Data-Driven Farming
The use of drone-assisted remote sensing is ushering in an era of 'precision agriculture'
Brian Luck grew up on an 800-acre corn and soybean farm in western Kentucky, so he knows well the look of a planted field from the exact height of a tractor seat.
But these days, Luck is more familiar with a much loftier view of farm fields. It’s a bird’s-eye perspective afforded by the “unmanned aircraft vehicles,” or drones, that have captured Luck’s imagination as an assistant professor of biological systems engineering and extension specialist in machinery systems at UW–Madison.
From a workshop in the Agricultural Engineering Laboratory, Luck has been working to wed the programmable flight of drones with the evolving science of remote sensing — imaging farm fields with spectroscopes and infrared cameras to reveal what the naked eye cannot see.
This summer, he and Shawn Steffan MS’97, an assistant professor of entomology will test knowledge gained from months of sweaty greenhouse studies by piloting their disease- and pest-seeking drone above cranberry bogs in northern Wisconsin. The work is being financed through a two-year grant from the Wisconsin Department of Agriculture, Trade and Consumer Protection and funding from the Wisconsin Cranberry Growers Association.
A better understanding of the data and images gathered by the drone-borne instruments could lead to new ways for cranberry growers to detect insects and disease weeks sooner than traditional scouting forays on the ground. Such foreknowledge would allow them to treat threatened plants earlier and avert more widespread damage and crop loss, according to Luck. And because farmers would know more precisely where to spray, they could reduce pesticide use, which would be a major cost saver and a boon to ecosystems already overburdened by chemicals.
It’s called “precision agriculture,” according to Luck and Steffan, who is also a research entomologist with the USDA Agricultural Research Service. And with all the work going on in their labs, greenhouses, and fields, drone-based precision ag is on the near horizon.
“The savings on inputs alone makes this work justifiable,” Luck says. “But until we show a farmer he or she is getting a dollars-and-cents benefit, until you show value, they’re not going to invest.”
THIS PUSH TO PROVE THE NEW TECHNOLOGY explains why labs such as Luck’s, along with other labs across the CALS campus, frequently look like the droid junkyards out of a Star Wars movie. Drones of every description and size perch on lab benches. Spare parts and controllers and snarls of coiled wiring crowd shelves and benches in the best mad scientist tradition.
With two fast-changing technologies at their fingertips — drones and remote sensing — it is sometimes difficult to tell which excites researchers the most.
Phil Townsend, a UW–Madison professor of forest and wildlife ecology, specializes in studying remote sensing. He is director of the UW–Madison Environmental Spectroscopy Laboratory, where research is pushing the science of remote sensing forward by finding new ways to incorporate spectroscopy and interpret data.
The science has come a long way, Townsend says. In the 1800s, scientists were strapping cameras to the legs of pigeons; and in the 1920s and 1930s, researchers used aerial photography to create the earliest maps of soil types.
Townsend’s lab specializes in reflectance spectroscopy, the study of how light interacts with objects. The instruments measure the intensity of light reflected by an object at wavelengths from 350 to 2,500 nanometers. This extends the reach of researchers far beyond the range of human eyes, which can only detect light across a meager 390 to 700 nanometers.
Using a type of spectroscopy called imaging spectroscopy, researchers are able to determine the chemical and physical composition of vegetation and other material based on the shape of the reflectance profile. These chemical signatures are what give researchers the ability to determine the health of crops over large areas. Simply put, a healthy plant is going to have a different chemical profile than a plant being devoured by caterpillars or destroyed by a fungus.
“We’re pretty good at knowing what we’re seeing,” Townsend says. “We take the spectral data and figure out what different molecules in the plant absorb light at different wavelengths. So we can say, for example, that an effect we’re seeing is because of a nitrogen deficiency.”
WHILE TOWNSEND CAN TALK nonstop about spectroscopy and its promise, Luck brings a similar energy to the subject of drones. Big and jovial and sporting red hair and beard, he roamed his lab one recent spring day, enthusing about piloting drones and the advancements that have put the technology at the fingertips of experts and non-experts alike.
Luck is enamored with all things mechanical. In fact, it’s nearly certain he would not object to being labeled somewhat of a geek about the subject. He’s spent a good part of his research career thinking about and studying how to improve the efficiency of all things mechanical on a farm. His publications deal with fine-tuning everything from agricultural sprayers to ventilation in chicken broiler houses. A favorite subject is autonomous farming, or, in plain English, robot tractors.
But few things cause Luck to light up as much as drones. Luck understood their potential value in agriculture a number of years ago and started toying with their inferior forerunners.
“I crashed a lot of cheap, remote-controlled airplanes,” Luck says.
Today, however, flying a drone is a lot less harrowing thanks to sophisticated onboard navigation systems and software that allow a pilot to program a flight plan and let the drone guide itself. Some software, for example, allows an operator to draw a line around the part of the field that needs inspection, and the software creates an automated flight path. The drones, especially those with horizontal, helicopter-like propellers, can hover over a specific location or fly low and slow, depending on the imaging task.
“I can tell the thing to fly 10 feet off the ground,” Luck says.
The drones used by Luck and others who are adapting them to agricultural uses are feather-light, made from carbon fiber. Black, four-legged, and bristling with antennae and other instruments, they look almost menacing, like crouching spiders. They’re controlled using an iPad that’s loaded with navigation software and linked to a controller equipped with dual joysticks.
Part of the confidence that Luck and Steffan have in the drones’ capabilities stems from a project initiated by Steffan a couple of years ago.
Moths are a particularly pesky problem for cranberry growers, and Steffan came up with an idea for dealing with them that could reduce pest populations while applying less insecticide. Instead of the standard insecticide regimen, Steffan used a sex pheromone that, when broadcast within a marsh, confused the male moths and kept them from mating.
“Reduced mating means fewer caterpillars chewing on cranberries,” Steffan says. “It’s moth birth control.”
The unique part of the plan involved deploying the pheromone with a drone. Steffan worked with a private company that mixed the pheromone into a “wax soup.” Then, using a novel contraption mounted to the drone, they dropped dollops of the cocktail (a trademarked substance known as SPLAT) into the cranberry bog. It worked. And the programmable, highly maneuverable drone made it possible.
“We basically showed it could be done,” Steffan says. “It’s a new form of precision ag for Wisconsin cranberries.”
The experiment was a perfect example of why drones are becoming a go-to alternative to planes and satellites for carrying remote-sensing instruments and other scientific payloads. They’re cheaper and more versatile, and they give the researcher more control over how and when to fly.
These attributes also bode well for getting the technology to farmers for their personal use in the near future, Luck says.
“I’m trying to investigate things farmers can get their hands on now,” Luck says. “This is something a farmer could buy tomorrow.”
Luck has no end of ideas when it comes to envisioning how drones can be incorporated into the routine operation of a farm. Farmers, he said, could use drone imaging to make management decisions about everything from fertilizer application to irrigation. Or, he mused, leaning against a lab bench, what about the potential advantages for farmers with disabilities?
“Say you have a broken motor on top of a grain bin,” Luck says. “You could avoid climbing up a 100-foot grain bin by flying a drone up to check on the problem instead of having to climb up a ladder.
“Or what about flying a fencerow? You could fly a drone close enough to a fence to see if the insulators are working.”
It’s not too far-fetched, Luck says, to think that, someday, having a drone on the farm will be little different than having a tractor parked in the machine shed. He envisions drones ensconced in boxes on the edges of fields, their batteries being recharged by the sun via solar arrays, waiting for the farmer to bring them to life and pilot them up and down their crop rows with the same nonchalance they bring to deploying something as prosaic as a manure spreader.
OF COURSE, LUCK ISN’T THE ONLY researcher out there with such an imagination. Drones are hot. Researchers in fields ranging from meteorology to wildlife ecology are finding ways to bring the drone’s unique advantages to bear on their work.
Drones have become so ubiquitous, and the technology has advanced so rapidly, that UW–Madison has rushed to firm up regulation of the devices to meet Federal Aviation Administration standards and to make sure that researchers and others don’t run afoul of the rules in their work. Last year, the university adopted a new drone-use policy that replaced a set of more restrictive rules. The policy includes an approval process and allows drone use for research and teaching.
Restrictions were necessary because of Madison’s urban setting, nearby airports, and the location of UW Health’s Med Flight base at University Hospital. The new regulations followed an unsettling 2014 incident in which a mysterious drone appeared and hovered above the Camp Randall Stadium student section during a football game. The pilot was never discovered, though the incident was heavily investigated due to the fear of a heavy drone dropping into a crowd of unsuspecting people.
Today, federal rules exist for both recreational and commercial use of drones. Farmers who use drones would have to abide by the commercial rules. A drone pilot certification is required, for example. Flights are restricted to daytime hours, altitudes limited to 400 feet and speeds to 100 mph, and pilots must maintain visual line of sight with their drone.
Also, as part of its larger response to the heightened interest in and use of drones, UW–Madison now offers a course on piloting the devices. Chris Johnson, director of the UW Flight Lab in the Department of Industrial and Systems Engineering, started the class last year. He said the class is aimed at not only teaching students to pilot drones but also introducing them to practical applications of the technology. The course includes a certification exam.
The list of commercial uses for drones is long and growing, Johnson says. Their use is being explored in construction, real estate, and utilities management. For example, Johnson says, utility companies could use drones to inspect thousands of miles of power lines, seeking hot spots or corrosion. But farmers especially are paying close attention to the adaptation of drones, he says.
“Agriculture is perhaps the single most mature market sector for drone-based technology,” Johnson says. He cites the cost savings for farmers and the greater ease and safety of flying drones in rural rather than urban areas as the major reasons for agriculture’s embrace of the technology.
“A lot of growers know about it,” says Joe Paul, a former farmer who runs a drone-based agriculture imaging and data service in New Lisbon, Wisconsin. “Although they’re not quite sure how they want to use it in their operations. It’s mostly the younger farmers, the sons or daughters who are taking over management and want to try new things. Drones will be able to help bring farming into a more technological age.”
One problem, Paul says, is that some farmers choose not to tackle the complexities of the software that is necessary for compiling and interpreting images and data gleaned from the drone flights. “A lot of growers don’t want to sit at the computer,” Paul says.
And that is where the evolving field of remote sensing technology becomes a crucial part of the story. As drones increase in sophistication, UW–Madison researchers are also developing a better understanding of how to read and interpret the images being captured by the drone-borne spectrometers and cameras. Equally important, they are finding ways to make that knowledge more accessible and usable for farmers.
“It’s basically automating the process,” says Townsend. “We’re out there figuring out how to manage this and turn it into data people can use.”
WORKING ALONGSIDE LUCK and Steffan on the cranberry project, postdoctoral research associate Jessica Drewry PhD’17 and lab technician Elissa Chasen PhD’14 have spent long months in the Walnut Street Greenhouse on the UW–Madison campus. They have carefully tended and grown cranberry plants before infesting them with an army of hungry caterpillars that munch away on the leaves.
Now Drewry and Chasen are meticulously documenting the damage the caterpillars have wrought using images taken with a multispectral camera, an infrared camera, and a regular camera. Like most such research, it has been a slow and tedious process — from building the wooden and mesh-covered frames that cover the plants to growing the cranberries to collecting hundreds of images week after week.
And, of course, there have been challenges. The researchers struggled to figure out a way to water the plants. At first they had the plants resting in trays while they watered them.
“But the caterpillars were falling off the plants and drowning in the trays,” Drewry says. They got rid of the trays and built a special contraption to water the plants without disturbing the caterpillars.
This is the hard, workaday part of science that most people don’t see. Entomologist Steffan calls it “ground truthing.”
Drewry and Chasen feed the data they collect from the images into a software program that helps them correlate the data from the pictures with the extent of the caterpillar damage. They do everything they can to take into account variables that would harm the accuracy of their findings. In the bright, steamy greenhouse room, for example, they noted temperature and light variations from enclosure to enclosure.
“That’s why we have a clipboard,” Drewry says. “And why there are two of us.”
At one point, the women bent over one of the plants, worried as parents hovering over a baby’s crib.
“We should note that fungus,” Drewry says. Down went a scribble on the clipboard. To keep their experiment as realistic as possible, they use the drone itself, outfitted with the cameras, to capture the images. Drewry holds the drone above the enclosures, and Chasen uses the drone controller to trigger the cameras.
This is the painstaking process that Luck and Steffan hope will eventually lead to a system that allows a farmer to easily analyze images of a field on a computer and understand immediately what’s revealed, whether it’s an early-stage caterpillar infestation or perhaps a fungus or other disease.
The brilliance of such a system is that the time-strapped farmer won’t need to learn a thing about the complicated science that powers the new technology. The farmer will sit down at a computer and load the images, and powerful algorithms — basically sequences of instructions built by the researchers using their collected data — will do the hard work of matching images with the specific nature of the crop damage.
“It’s basically automating it,” says Townsend. “We’re working underneath the hood, making all of this invisible to users.” This invisible science of algorithms powers much of the technological innovation we rely on today, from weather maps to navigation devices to Google searches.
OTHER SCIENTISTS IN CALS ARE also working to expand our remote sensing capabilities.
In the environmental spectroscopy laboratory, researchers Clayton Kingdon and Erin Hokanson Wagner work with Townsend to create new uses for spectrometers as well as a more refined understanding of the data collected by the instruments. And they are compiling a database, basically a library, of spectral data called the Ecological Spectral Information System, or EcoSIS.
Spectrometers of every stripe sit on the lab benches in various stages of assembly. On one countertop is a tube-shaped spectrometer mounted on what looks like a trailer hitch. It’s designed to be mounted on a tractor.
Open just about any drawer in the lab and you’ll find a spectrometer. The whiteboards in the lab are covered with dozens of scientific scribbles, various versions of the distinctive squiggly lines that represent light at different wavelengths, all created as the researchers hash out the data they are continually compiling and interpreting.
Kingdon is particularly enthralled with dreaming up different ways to deploy the spectrometers, including on airplanes and on two different kinds of drones that are at rest on a counter in another room.
In March 2018, Townsend spent time at the NASA Jet Propulsion Laboratory in California, working with researchers there to adapt the newer, cutting-edge hyperspectral imagers to use aboard satellites. Surprisingly, he says, few, if any, satellites are equipped with the newer versions of imaging spectrometers now being pioneered.
For Townsend and the others in the spectroscopy lab, the most exciting work involves the development and use of hyperspectral remote sensing, an amped-up version of traditional spectroscopy. It provides much more detailed data, offering the promise of detecting problems with a crop even earlier.
Like Luck and his crew, Kingdon and Wagner are working with farmers to compile additional data from field studies to bring more precision to the use of spectroscopy. They’re working with cranberry and potato growers and others. But the work in the lab goes beyond agriculture to studies, for example, of forest canopies in northern Wisconsin looking to better understand the impacts of climate change.
Wagner has been working on a project in which she is using spectroscopy to better predict yield — crucial for growers who are trying to market a crop even as it grows in their fields.
All of this science has moved agriculture far from the days when a farmer walked out in a field and cupped a handful of soil to make management decisions. And it has propelled Luck into a much different domain than those rural Kentucky corn and soybean fields. He still honors that past. He still feels happy and at home in the seat of a tractor. But when Luck starts talking about drones and the promise of remarkable machines that can peer into worlds invisible to our eyes, it’s the future one hears.
“There are uses out there for this technology that we don’t even know about yet,” Luck says.This article was posted in Cover Story, Features, Food Systems, Main feature, Summer 2018 and tagged biological systems engineering, Brian Luck, Clayton Kingdon, cranberries, drones, Elissa Chasen, Entomology, Erin Hokanson Wagner, Forest and Wildlife Ecology, Jessica Drewry, Phil Townsend, precision agriculture, remote sensing, Shawn Steffans, spectroscopy.