Nothing Runs Like a Precision Farming System
John Deere engineers apply high technology such as GPS to farming, improving efficiency and reducing costs and environmental impact
By Jill Brimeyer
In 1994, top engineers from agricultural equipment maker John Deere found themselves huddled around a conference room table. The mission: seek out the future technologies Deere should pursue to further sharpen the company’s competitive edge. From the outset, Terry Pickett, manager of the John Deere Ag Management Solutions (AMS) Advanced Engineering group in Urbandale, Iowa, knew what was coming. Many more days would be spent in the conference room at Deere and Company headquarters in Moline, Illinois tracking scribbles on whiteboards and volleying ideas. Some were more promising than others, but a strong theme emerged.
“When we whittled our ideas down, precision agriculture kept bubbling up,” recalls Pickett, who bills himself as an electrical engineer who likes to play with agriculture. “We ended up convincing Hans Becherer (then Deere’s chairman and CEO) that pursuing this market was a good idea.” Precision agriculture, which uses signals from the constellation of Global Positioning System (GPS) satellites for farming applications, showed promise. But at the time, there were far more questions than answers.
Competitors had already developed yield mapping systems for combines that measured the mass flow of grain by using an impact plate at the top of a grain elevator. Data was collected instantly and paired statistically to the plot of field that produced it. With this information, a farmer could determine if his 130 bushels per acre is due to some areas yielding 180 an acre and others only 80 and treat those fields accordingly. Despite the idea’s promise, most companies that offered the technology floundered. It was difficult to attain the necessary level of GPS accuracy needed for these farming applications without driving up cost. And it was even harder to convince farmers of the systems’ payback potential.
“If you go back to 1994, ’95 and ’96 and look at precision farming, the emphasis was on yield mapping and variable rate application,” recalls Fred Nelson, a principal engineer at John Deere AMS and another member of the original engineering team. “Sure, they could produce lots of pretty yield maps, but what could you do with them?” Nelson says creating a colorful record of a field, showing which areas produced the most and which needed more seed or chemicals, was helpful but just not enough. “Some growers could use them, but beyond that, most farmers found that the cost couldn’t be justified without good agronomic understanding of cause and effect. So at the early adopter stage, the technology never got to most users.”
Solving the GPS Puzzle
In 1997, as less hardy competitors began to sink and leave the market, John Deere engineers started in earnest to tackle a mighty task: solving the technical “impossibilities” that prevented precision ag from becoming fully viable and profitable. As it turned out, a small project with Stanford University and Bob Mayfield, an engineer with the John Deere Advanced Tractor Engineering group in Waterloo, Iowa, came to focus. This project looked at the possibility of automatically guiding a tractor in the field. The technology to take GPS accuracy down to the 2-to-4 centimeter level existed, but it took two Ph.Ds to support it due to its sheer complexity.
It became clear the linchpin of a successful guidance system would be a simple GPS system supported by the John Deere dealer network. So while Mayfield continued to work with Stanford engineers to refine the overall product concept, Nelson, Pickett, and their Advanced Engineering peers went back to the drawing board. Teamed with talented engineers from NavCom Technology, now a wholly owned subsidiary of John Deere, and NASA’s Jet Propulsion Laboratory, the group spent weeks deconstructing problems and looking at new ways to put the pieces back together. One of these challenges was differential correction, the ability to fine-tune GPS signals by using a ground receiver to adjust for error. Without accurate correction, readings could be off by half the length of a football field.
In 1998, GPS accuracy of mere centimeters was pie in the sky. But for this group of engineers, their fresh eyes worked to their advantage. “People who develop GPS have a paradigm,” explains Nelson. “Those of us who didn’t grow up in the GPS world don’t have those paradigms. You can more easily shift into a different direction and make a dramatic difference.”
Pickett does admit that not having experience in GPS also worked against his John Deere team in many respects. They weren’t entirely familiar with the variety of GPS frequencies, and myriad other technical puzzles needed to be solved. A key factors in most of the solutions, the John Deere team says, was that the GPS experts at NavCom were willing to give their ideas serious consideration. “I don’t know how many times we ended up in a room at a standstill, saying there’s no way something could be done,” recalls Pickett. “Then we went back, tore it apart piece by piece, and found a way to do it.”
The team’s foremost task was to tweak the capabilities of differential GPS (DGPS) correction to attain the accuracy needed for guidance applications. Competitive DGPS systems were designed to work in regional areas with a network of base stations, each station computing its correction. By broadcasting these corrections, GPS units in the field could figure out which station it was closest to and correct its position based on that data. An inexact science, it worked for the precision ag applications available at the time.
To reach its accuracy goals, Deere created its own means for differential correction. “We came up with the idea to create a set of corrections for each satellite, rather than a correction for each base station,” says Nelson. “There’s one for each satellite, orbiting anywhere in the world, and then one for each region. It’s a cheaper system to run, and it improves accuracy.”
The result of their efforts can pinpoint location within an inch and is billed as the first truly global DGPS network as well as the most accurate and reliable system available. “Our collaborations allowed us to extend DGPS to the entire globe, with a signal valid anywhere you go in the world,” states Nelson. Pickett shakes his head at the memory. “We really turned around precision ag to a level that no one had even dreamed about.”
The next hurdle was determining how to isolate and shield the receiver’s sensitive circuits in a cost-effective, reliable way. The engineering team discovered that to save money, they needed to invest more in the beginning. “We’re not averse to using a $60,000 tool, or a $100,000 tool, to get the job done right. Other GPS people didn’t understand this,” says Nelson. “They would create packaging that would rack up a higher unit cost and lower reliability. But we knew that if you came up with a new concept, a new system, you would have better luck in the field.” As a result, the team spent more money initially to create modular components with integrated packaging, rather than putting unprotected components together and then throwing even more money at them to try to insulate them down the road.
Many in the industry questioned John Deere’s approach as it set out to reinvent everything right down to their own set of application-specific integrated circuits, chipsets, and controls. “People may have asked, ‘why would John Deere build their own GPS when they could acquire the technology so easily?’” remarks Nelson with a chuckle. “Well, if it’s something we build ourselves, it’s repeatable year after year. We couldn’t guarantee that if we used another manufacturer, particularly when we were only a small part of the growing commercial GPS industry.”
Farming the High-tech Way
The John Deere engineering team solved its technical riddles, and the results proved dramatic. In 1998, the company first offered satellite-based DGPS systems with one to two meters accuracy. By 2004, accuracy had increased to 10 centimeters, and today the numbers continue to shrink. Deere offers its StarFire receivers with several levels of differential correction, so farmers only need to pay for as much as they need for the job – lower accuracy for field mapping, higher accuracies for automatic steering applications. Now, even tractors for sale in the used market can be retrofitted with new technology.
Once Deere had ironclad GPS in place, the company exploited it through several series of modular, interoperable devices for the farm. Some of the most “gee whiz” items are those that enable tractors, combines, and similar machines to virtually steer themselves. GreenStar Parallel Tracking guides the operator through the field, using a digital line on a display to guide the farmer’s straight or curved steering. The next step up, GreenStar AutoTrac Assisted Steering, is a hands-free system in which the equipment locks into and guides itself along straight rows.
“If a farmer’s machinery typically has 10 percent overlap in the field, that’s three feet of overlap using a 30-foot implement,” explains Nelson. “If you give accurate guidance, that’s 10 percent less time in the field, chemicals, and wages. When you calculate that, it’s very easy to justify the steering system. And that’s not even counting being able to farm in conditions when you couldn’t have worked before, like in darkness. Some years, that makes the difference between having a crop and not having one.”
John Deere also took yield mapping and cranked it up a notch, extending its Combine Yield Mapping System to cotton pickers. This variation of the system used for grain leverages Doppler radar technology to measure the mass flow rate of the problematic fluffy stuff. It beams microwaves, which illuminate the stream of cotton and then reflect to a sensor. The ability to accurately measure yield is changing the face of farming. “In the early stages of precision farming, the variable rate information never could be conclusive,” says Pickett. “You’d have to base your current year’s yield on last year, which doesn’t do much good. It all relies on nitrogen availability, which varies greatly. Now it can be a real time, in-season activity that puts on nitrogen when the crop can use it, improving the environment while going easier on the producer’s wallet.”
John Deere AMS has come a long way from 1997 and that seed of an idea planted in an Illinois conference room. Even so, the process, and progress, continues. The precision ag concept will be implemented on a wider range of vehicles in the future, and studies are continuing on fully autonomous vehicles. Right now, the latter is looking like another one of those “impossible problems.” But don’t write off the idea just yet — the engineers at John Deere seem to specialize in impossibilities.
“The best meetings we’ve had were like a thundercloud — utter chaos,” muses Pickett with a smile. “You start by saying, ‘how are you going to get around this?’ Then discussion builds up and builds up, and at the end of the day you come up with a solution that was so unique, you are just amazed with it.”
For more information on John Deere Ag Management Solutions, visit www.deere.com/en_US/ag/servicesupport/ams/index.html
Jill Brimeyer is a freelance writer in Ankeny, Iowa
Armed with artificial intelligence, robots are working in recycling facilities to address contamination, safety, and manpower issues
Headquartered in Oak Ridge, Tennessee, engineering firm Process Engineering Associates specializes in one discipline, but they apply it to many types of projects all over the world
Engineers at the FREEDM Systems Center at North Carolina State University are developing solid-state transformers that promise to make the electrical grid more reliable and facilitate renewable energy such as wind and solar
Old dams are being taken down around the country for environmental and safety reasons. In Massachusetts, the story of the Upper Roberts Meadow Reservoir Dam removal project shows the complexities involved and the opportunities for engineers.
Mechanical engineers help with knee injuries by developing computational models to characterize ligaments.