Spring 2019 Vol. 6, No. 1
Research explores ways to improve GPS accuracy for traffic safety applications
Lane-departure or lane-keeping systems, which alert distracted drivers when they’re straying from their lanes, have the potential to greatly improve driver safety. A 2017 study found that they reduce crash and injury rates by 11 and 21 percent, respectively. However, the Global Positioning System (GPS) technology needed to make them widespread has yet to be perfected.
A lane-keeping system needs to have an extremely accurate GPS—down to 2 to 4 inches. But those systems can cost thousands of dollars and must be installed in the factory. For lane-keeping systems to be truly useful to the wider public, they need to be less expensive and easier to install after market.
Rhonda Franklin
One way to make these systems more affordable would be to use low-cost GPS antennas; however, along with their reduced cost comes reduced accuracy. In a Roadway Safety Institute project, researchers are seeking to solve questions: why are high-quality antennas so accurate, and can antennas be calibrated outside of the lab?
“If we could determine the errors and then resolve those errors, it could potentially open the door for using low-cost devices like your cell phone,” says Rhonda Franklin, principal investigator of the project and a professor in the University of Minnesota’s Department of Electrical and Computer Engineering (ECE).
Granted, the RSI-funded study is preliminary and the possibility of using a cell phone antenna for a lane-keeping system is a long way off, Franklin says. However, she and fellow researchers Robert Sainati from ECE, Demoz Gebre-Egziabher from the Department of Aerospace Engineering and Mechanics, and a team of students decided to tackle some emergent questions: why are high-quality antennas so accurate, and can antennas can be calibrated outside of the lab?
Franklin led a team that took two GPS antennas—one high-cost and the other low—and put them in the controlled environment of an anechoic (echo-free) chamber to test how they “see” satellite signals. From this, the researchers gained an understanding of the antennas’ “phase center”—the point where satellite signals hit the antenna. They found that this point has a tendency to wander, which causes GPS readings to be off by a few millimeters or even centimeters. This can cause large errors in calculating position since the signals from multiple satellites are not hitting the same point on the antenna.
Testing anechoic chamber used to characterize the phase center at
different elevation angles of the satellite. The green line depicts the
turntable axes of rotation relative to the calibration horn. In this
image, the test antenna is removed.
Gebre-Egziabher tackled the second part of the problem: calibrating an antenna outside of an anechoic chamber. After collecting 48 hours’ worth of data from high- and low-cost GPS antennas placed outside, his team set about trying to create error models that could calibrate the antennas to the needed level of accuracy.
“If you could do that, people wouldn’t have to spend a lot of money on calibrating antennas in expensive anechoic chambers,” Gebre-Egziabher says.
The ultimate goal of this work is to make lane-keeping systems available to more people. If low-cost GPS antennas could be made as accurate as high-cost antennas without built-in calibration, these systems could be mass-produced and sold as an add-on to retrofit existing cars. More people would be likely to buy them and benefit from their safety features.
Getting to this point, however, will take more work. The researchers now know that the calibrations can be done, but the data-collection apparatus they used had some flaws. With more funding to continue its effort, the team would set up a new experimental apparatus and collect additional data, Gebre-Egziabher adds.