Drone's Eye View
If you do the math, farmers don’t make much money,” says Drew Wilkerson, Ph.D., Assistant Professor of Mechanical Engineering at York College. He says a typical acre can yield maybe 200 bushels of corn on the very best of days, but often it is far less. In today’s market, a bushel of corn fetches about $3.70.
“That’s $740 an acre, but if you subtract out the cost of seed, herbicide, pesticide, and fertilizer and a typical 100-acre farm is lucky to clear $30,000 a year,” Wilkerson explains. “Many farmers have to have part-time jobs just to stay in business.”
It may seem odd that a mechanical engineering professor would know so much about agriculture, but Wilkerson has a unique perspective on the subject — literally. He has designed a senior-year capstone course for York engineering students where they build, program, and operate cutting-edge drones equipped with specialized cameras to help farmers become more efficient. He says they have had tremendous support from the local community — in particular Dave Shepp, Ryan Heindel, and Bill Whale of the York Area RC Club.
Wilkerson and his students photograph York County’s farms and analyze the light reflecting off the plants below. From this rich data, using sophisticated computer applications, they can assess how well the plants are doing, where fertilizer is most needed, and whether insects are taking a toll. The farmers can then use that data to optimize their chemical use, spraying expensive — and environmentally consequential — fertilizers, herbicides, and pesticides only where they are most needed.
“We are trying to reduce the amount of the chemicals while keeping the yields high, but there’s a lot to learn for the kids, too,” Wilkerson says.
The drone is designed, built, and operated by students. Guided from the ground, it soars high above the neat rows of crops, zipping back and forth in a tight lawnmower-like pattern across the fields. Each pass is closely controlled by GPS.
The drone is fitted with a specialized camera that has no fewer than five distinct lenses, each capturing different types of light being reflected from the crops below.
Wilkerson runs down the litany of light with the fervor of a true engineer — it does near-infrared, red-edge, red, blue and green, and invisible light, he says.
“The camera alone is $3,500,” Wilkerson notes. “That’s a lot for a small group like ours, but it makes for a fun senior course that touches on lots of areas of engineering.”
Once the photography is captured and downloaded, the images are stitched together photographically, the way Google creates bird’s-eye views of the world from multiple satellite images. The result is a mosaic — an entire farm drawn in bright primary colors. While those images are not necessarily beautiful to the untrained eye, in the hands of experts they mean the world. Hidden in the light are untold secrets that can help farmers optimize their investments. This is known in technical terms as multispectral analysis.
“All this knowledge from the light reflected off of the plants. It’s pretty incredible,” Wilkerson says. “You can photograph a 100-acre farm in about 15 minutes.”
Wilkerson also notes saving money is one of many benefits of this technology. There are environmental upsides, too. Fertilizers like nitrogen and phosphorous can leech into the water supply and cause harmful algae blooms that wreak havoc on aquatic ecosystems by sucking oxygen out of the water and creating so-called “dead zones” where no creature can survive.
The typical corn farm receives two doses of nitrogen per season, once as seeds are sown in the ground and again when the corn is about two feet tall. Less spraying means lower costs for the farmer, but also less nitrogen in the water supply.
While multispectral analysis may pay dividends in the wallet and on the waterfront, it helps in other more subtle ways, too. For instance, Wilkerson adds, agricultural experts and entomologists have become alarmed in recent years by the decline of bee populations. Bees are critical to agriculture for distributing pollen and aiding plant reproduction. That decline has been correlated with the use of certain insecticides, known as neonicotinoids, that are often sprayed liberally on fields. Helping to reduce the amount of neonicotinoids used or by increasing the precision of the spraying could have beneficial effects for bee populations.
One of the indicators of nitrogen is a light index called NDVI — short for normalized difference vegetation index. In technical terms, NDVI is a ratio of near-infrared light and what is known as red-edge light. While the technical details can be challenging for non-engineers, the results are obvious to anyone with just a glance at one of Wilkerson’s multispectral images.
“All that reflectivity gives us a weird-looking plot,” Wilkerson says. “But, when you look at it, where you see green, that means there’s nitrogen and that’s good news. No need to spray there. If it’s red that means a deficiency and you want to spray in those spots.”
While it seems remarkable that all this insight can be drawn out of the light, what follows next is even more remarkable. Wilkerson can download his spectral analyses, complete with GPS waypoints of spray and no-spray areas that are spatially accurate to 10 centimeters. This data all fits on a thumb drive which he then plugs into a GPS-controlled smart tractor and computers do the rest. Carefully guided by GPS, the tractor is able to precisely control the application of chemicals.
“The tractor knows when to spray and how much to spray and it reduces the amount that the farmer has to put on the crop. This saves them real money,” Wilkerson says.
Wilkerson got his start as a military research scientist where he worked for the U.S. Army for more than 33 years before joining the faculty at York College two years ago. He earned his Ph.D. at Johns Hopkins in 1990 and is an expert in robotics and unmanned systems. For the last 15 years of his military career, he focused on robotics and drones.
When he first started at York, Wilkerson joined a drone flying club, as much as a hobby as anything else, he says, but then he saw a video of a team of researchers using a
drone to analyze the canopies of the tall Sequoia forests out west. He saw a niche.
“I pitched it to the school and that’s how we got the capstone class. It’s been a lot of fun,” Wilkerson says.
A Team of Teams
The capstone team of seniors is actually a collection of four smaller teams that each takes on certain engineering responsibilities. There is plenty of work to go around and need for almost every sort of engineering talent out there, mechanical, chemical, aeronautical, computer science and so forth.
First, there is a team dedicated exclusively to learning to fly the drone. The Federal Aviation Administration (FAA) requires drone pilots to have a license.
“We call it flight ops,” he says. “The goal is to earn what is known as ‘Part 107’ FAA license.”
This is no easy task. There is a security clearance as well as subject matter exam. Last semester, just one of Wilkerson’s students passed the exam. This semester he thinks two, possibly three, others might earn their wings.
“And they have to study a lot. The sectional charts. They have to know weather. They have to know lots about drone maintenance, too,” he says.
Another team is known as “the farming group.” These team members are charged with learning everything they can about the business of farming, so that the knowledge imparted by multispectral analysis can be put to good use.
In short, the farming group has to learn about growing corn and soy bean, including what pests attack corn and what nutrients corn needs, as well as how pH of the soil affects yield. It is a multidisciplinary team. Wilkerson says he will probably have one mechanical engineer on the farming team and several computer science majors.
“They really have their work cut out for them. They need to know all the different tests that can be done using multispectral imaging to assess those characteristics of a field of crops,” he says.
The third group is an airframe team that designs and builds the drone. And, when Wilkerson says “design,” he means it. The drone must be able to stay aloft for an hour, enough to photograph up to 500 acres at a time. While this parameter is far larger than most farms in York County, it is still much smaller than the biggest plots, which can be 2,500 acres or more.
“They can’t just go buy some drone off the shelf. They’ve got to create it. We’re using a flying wing design and they need to know why that design is so efficient and good for these purposes,” he says.
Last but not least, there is the autonomous control group that programs the autopilot that guides the drone as it makes its finely tuned flight patterns in the sky. They are responsible for recovering the airplane if there’s a problem, as well. No minor duty when the drone and its expensive camera hang in the balance.
At the conclusion of the yearlong capstone course, the students make a group report and presentation of their findings before earning their diplomas and heading off to their next engineering adventures. This year’s presentation marked the culmination of the first year of Wilkerson’s capstone course. It was a year of tremendous learning opportunities and personal growth for the students and himself, Wilkerson says.
“It was a great first year, but there’s a lot still to work on,” he says. “I’m looking forward to next year already.”