Centimeter-Level Accuracy — Is It Coming?

ION 2007 News Coverage:
Jeff Chappell Blogs

Plenary Report:

What is the future of civilian GNSS?

Cars that drive and navigate by themselves, with the assistance of on-board sensors and GPS? Accuracy well below one meter–perhaps even the centimeter level, and all that implies–thanks to the promulgation of new civil signals from GPS, GLONASS, Galileo, etc.? Without the promulgation of post processing and other tricks used to improve accuracy today?

GPS World“The hopeful, qualified answer is yes, say speakers here in Ft. Worth, Texas at the 20th International Technical Meeting of the Institute of Navigation’s Satellite Division—2007 ION GNSS for short. But that bright future is dimmed somewhat by the problems and challenges of crowded orbits and crowded frequencies.

As far as self-driving cars are concerned, that future has already begun. ION’s Tuesday evening plenary session opened with keynote speaker Sebastian Thrun. Thrun, a professor at Stanford University, headed up the team of students that won the 2005 DARPA Grand Challenge, creating an autonomous vehicle that navigated 132 miles of Southwest desert terrain in 6 hours and 53 minutes.

Thrun and his students experience is a classic example of the capabilities and limitations of GPS technology when it comes to terrestrial navigation. In the 2004 Grand Challenge, every team’s autonomous vehicle failed in the first few miles, because of an over-reliance on GPS.

“A lot of people thought it was easy. I have a car, I have a computer, I have a GPS system,” Thrun recalled of the 2004 contest, which involved a course covering 2,700 GPS-based waypoints. As anyone who has depended solely on a PND or handheld GPS unit knows, the problem is that GPS data is not always accurate; particularly in terms of exact precision, there can be errors to contend with. Furthermore, it doesn’t account for obstacles and variations in terrain.

The answer for Thrun’s team was to put a suite of laser sensors on the roof of their vehicle constantly scanning the terrain ahead of the vehicle, creating a 3D cloud of data points that described the shape of the nearby terrain. To overcome the short-range effectiveness of laser sensors, Stanford’s winning system also used digital cameras for long-range terrain perception. It would create a terrain model based on the short-range laser data and apply that model to the long-range image to predict the long range path, constantly updating it as new short-range data became available.

In essence, the answer for an autonomous vehicle was relative positioning, not the absolute positioning of GNSS. Of course, the vehicle still depended on GPS data to describe its intended path. As Thrun noted, to actually determine where one currently is, one still needs the absolute positioning of GPS.

The next step for Thrun’s Stanford team and others answering DARPA’s call for autonomous vehicle technology is the 2007 Urban Challenge—a 16-mile course in an urban setting, complete with traffic. While that contest will take place later this year, Thrun discussed what the Stanford team is doing—namely updating its system with a 360 degree infrared sensor, supplemented with digital map data.

Is Centimeter-Level Accuracy Coming?

So what’s the one thing the GNSS industry could do to help make autonomous vehicles commonplace? Something that’s been talked about for decades? Thrun’s short answer is 10cm accuracy—it was the lack of extreme accuracy in GPS coupled with dependency on it that caused the entire field of competitors in the 2004 Grand Challenge field to fail, he reminded the audience.

Jeff Chappell at ION on behalf of GPS World. (where we didn't have a big photography budget)
“If you could do that, it would change the world,” Thrun said of centimeter-level GNSS accuracy. “Its impact would be fundamental to all of us.”

Several plenary speakers suggested during a question and answer session following Thrun’s speech that this level of accuracy might be closer than we think. Given the multitude of civil signals that will be available in the future, between the new signals being added to GPS, coupled with the availability of other civil signals from GLONASS and potentially Galileo and China’s Compass, centimeter-level positioning accuracy may be obtainable, said Günter Hein, of the Institute of Geodesy and Navigation at the University FAF in Munich, Germany.

The challenge, suggested ION fellow Gaylord Green, is capturing and ensuring the accuracy of all of the necessary satellite ephemeris data. “But it is coming,” he agreed. Green, incidentally, as an officer in the U.S. Air Force—now retired—was a member of the original Air Force team that developed and implemented the GPS system.

And as Hein and others have pointed out here, all those civil signals on the same frequencies are going to make things rather crowded, which could prove problematic in the future.

While it may seem to simply be a matter of getting enough space craft in orbit and coordinating the data they provide to achieve the goal of centimeter-level accuracy in GNSS, that’s actually no small or simple matter in reality. As U.S. Air Force Major General William Shelton informed the ION plenary session audience, for one thing, GNSS costs money, and lots of it. To date the United States has invested about $32 billion in the Global Positioning System; it costs about about a billion dollars a year to administer and maintain it, he noted.

Shelton is Commander of the 14th Air Force (Air Forces Strategic-Space), Air Force Space Command, as well as the Commander of the Joint Functional Component Command for Space, U.S. Strategic Command at Vandenberg Air Force Base, Calif.

While GNSS and GPS in particularly may have been driven early on by military applications, GPS was designed from the beginning to be a dual-use system, Shelton observed, and that it is civilian commercial applications that have become a principal driver of the technology. “The commercial aspect of space is what is absolutely exploding right now,” he acknowledged.

But this introduces a question of policy: just who protects commercial interests in space, and who will act as a traffic cop, keeping the increasingly-crowded geosynchronous and near-earth orbits orderly—and keeping in mind geo-political situations on the ground. “That’s a big policy question for us,” Shelton said.

He suggested that it was going to take collaborative efforts of everyone involved in the military and the private commercial sector to protect the interests of all the participants in space, namely GNSS. “We believe this is the policy of the future.”

Editor’s Note: As explained at length elsewhere on this site, this is a news story/blog entry written by me that originally appeared on GPS World. GPS World and parent Questex Media hold all of the rights. You can still see a copy of this story at GPS World.