CONTACT: David F. Salisbury, News Service (650) 725-1944;
Future farmers of America may never have to learn to drive a tractor.
A Stanford research team has equipped a John Deere tractor with a satellite-based automatic control system that can guide the 20,000-pound farm vehicle more precisely than the best human drivers.
Doctoral students Gabriel H. Elkaim and Michael O'Connor report that their team has perfected the system to the point that it can control the tractor at all speeds while pulling a variety of implements, with centimeter-level precision.
Courtesy of GPS Laboratory
Members of the Stanford GPS research team posing in front of the John Deere tractor that they have used to develop a satellite-based automatic control system.
The team, which is supervised by Bradford Parkinson, the Edward C. Wells Professor of Aeronautics and Astronautics, previously had reported obtaining this level of precision at specific speeds and loads following both straight-line and closed loop paths . The students presented the new findings Sept. 19 at the Institute of Navigation's GPS-97 conference in Kansas City.
"These guys have done an amazing job in controlling the tractor. They've taken it beyond anything we expected," said Gary K. Hendrickson, manager of product test and reliability for John Deere, who has been working with the Stanford team.
The research group achieved this unprecedented level of control by using the Global Positioning System (GPS), a constellation of 24 satellites operated by the U.S. Department of Defense. The satellites circle the Earth every 12 hours. The system originally was designed and constructed primarily for military purposes, but it has been made available for civilian use.
Normal civilian GPS receivers have a precision of about 100 yards; a system called differential GPS, which requires a local base station, can provide meter-level accuracy. At these accuracy levels the technology has found widespread use in the nation's farmland, as part of a movement called precision agriculture.
In the days before agriculture was industrialized, farmers were familiar with the characteristics of their fields yard by yard, and successful farmers adapted their practices, such as seeding, cultivating and irrigation, to these variations. The advent of industrialization, however, forced farmers to prepare entire fields in the same way, regardless of the variations in soil conditions and other factors.
By mapping their fields with GPS, innovative farmers now have begun to adapt their seeding and fertilization practices to take these small-scale variations into account once again. Using these electronic maps, for example, farmers have begun varying the amount of seed they plant and the fertilizer they apply. In some cases, they are doing these tasks manually, guided by an inexpensive handheld GPS receiver. In others, they are using GPS-guided systems that automatically vary the application rates.
Although these efforts remain experimental, proponents argue that precision agriculture will improve farm productivity while reducing applications of fertilizers, herbicides and pesticides. Not only would this save farmers money, it also could reduce the adverse environmental impact of their operations. Some proponents even are calling these developments the "third agricultural revolution."
Until now, however, the degree of precision available from GPS has been inadequate for control of farm vehicles. To achieve this control, the Stanford researchers adapted a method that they had developed for an automatic aircraft landing system in 1994. The system employs an additional ground station, called a pseudolite, that puts out GPS-like signals. When combined with differential GPS, the system provides information on the tractor's position and attitude with centimeter-level precision.
Automatic control of ground vehicles has been a research objective for many years. Such control has many potential applications, including development of smart roads that automatically pilot vehicles to the destinations their users select; control of construction vehicles that build roads automatically; and control of vehicles operating in hazardous environments.
"We think that farm vehicles will be the first major application," said graduate student O'Connor, citing several reasons: Solid objects such as mountains and tall buildings block the satellites' signals, so a large expanse of sky must be available for the GPS system to work properly, which is normally the case in agricultural fields. Farmers and farm equipment manufacturers already are embracing GPS technology, O'Connor said, and the potential economic benefits to farmers are large enough to provide a real incentive for its adoption.
Although the ultimate application of this technology would be an autonomous tractor that a farmer can command and monitor from the office, there also are potential advantages to adding such a control system to tractors with drivers. Bedding, seeding, disking, fertilizing and similar procedures are among the most monotonous and time-consuming tasks that farmers face. Many farmers tell stories about falling asleep on the tractor and taking out a couple of rows of crops.
"This kind of control system, which does the basic driving, frees the operator to make higher-level decisions," O'Connor said. It also could allow less experienced operators to plow more precisely than veterans. The accuracy of the best human tractor drivers that the Stanford researchers have been able to measure is about 10 centimeters. That compares to a 2-centimeter error with the satellite controller.
Such high level of precision is not required for all field preparation, but it is necessary for some applications. In growing melons, for instance, farmers use buried hoses, called tapes, to water the crops. Once a field has grown up, however, it is difficult for farmers to locate the buried tapes. Typically, they dig up each end of the row to locate the end of the tapes and then mark them with flags. The tractor operator is expected to drive the entire length of the field and avoid the tapes by sighting on the distant flags. Despite their best efforts, the tractor operators do thousands of dollars' damage to the tapes annually. With the GPS control system, by contrast, the exact location of the tapes could be recorded at the time they are installed and the tractors programmed to avoid them.
Such a system also would allow farmers to operate during the night, through heavy dust or fog. This capability could potentially minimize losses at times when crops must be harvested as quickly as possible.
Satellite-based controllers also might allow farmers to do things that are impossible or impractical manually, such as enabling a single driver to operate a convoy of several tractors at the same time. Or it could allow farmers to plow their fields in a spiral pattern, rather than in parallel rows. A spiral pattern would allow a farmer to plow a large field continuously, without wasting time making U-turns at the end of each set of rows, but it is very difficult for drivers to do.
The Stanford project began four years ago when O'Connor asked Parkinson if he could try to develop a control system for an electric golf cart. In a year, O'Connor and two other students had a working system. At that point, the Stanford researchers approached the John Deere Co. for support. The company supplied an older, mid-sized tractor model, the 7800, which weighs about 20,000 pounds, has a 140 horsepower engine and can travel at speeds up to 22 mph.
Since then, the team has grown to seven students. In addition to O'Connor and Elkaim, doctoral students Tom Bell, Andy Rekow and Ajit Chaudhari work on various aspects of the system. Two undergraduates, Arti Garg and Seebany Datta-Barua, work on related topics.
O'Connor, who will graduate in a few months, said he hopes to get a job building a commercial prototype of the tractor control system.
The research has been supported by Deere and Company. It is a spin-off of previous Stanford research that was sponsored by the Federal Aviation Administration and the National Aeronautics and Space Administration. Trimble Navigation provided the GPS receivers used in the project.
For more information, see the GPS Laboratory home page under research projects, land vehicles. The URL is http://www.stanford.edu/group/GPS/.
By David F. Salisbury