A New Technique Makes GPS Accurate To A Centimetre

A New Technique Makes GPS Accurate to an Inch

GPS is an utterly pervasive and wonderful technology, but it's increasingly not accurate enough for modern demands. Now a team of researchers can make it accurate right down to a centimetre. Regular GPS registers your location and velocity by measuring the time it takes to receive signals from four or more satellites, that were sent into space by the military. Alone, it can tell you where you are to within 9m. More recently a technique called Differential GPS (DGPS) improved on that resolution by adding ground-based reference stations -- increasing accuracy to within 1m.

Now, a team from the University of California, Riverside, has developed a technique that augments the regular GPS data with on-board inertial measurements from a sensor. Actually, that's been tried before, but in the past it's required large computers to combine the two data streams, rendering it ineffective for use in cars or mobile devices. Instead what the University of California team has done is create a set of new algorithms which, it claims, reduce the complexity of the calculation by several order of magnitude.

In turn, that allows GPS systems in a mobile device to calculate position with an accuracy of just a centimetre. The research is published in IEEE Transactions on Control Systems Technology.

The team hopes that the new GPS units could be used where accuracy is far more important that it was in the past. Autonomous vehicles is an obvious application, where knowing exactly where the vehicle is on the road is absolutely crucial -- but it could be included in your phone, too.

[IEEE Transactions on Control Systems Technology via PhysOrg]

TechnImage by Aaron Parecki under Creative Commons licence



    I was under the impression that GPS was always much more accurate, it's just the military purposely make it less accurate for civilian use. Maybe its about time the military simply abandon the technology that makes it less accurate since it's effectively obsolete.

      Surveyors RTK's have an accuracy of 10mm (using a base station).

      yeah, I thought the only thing holding GPS back from being crazy accurate was the sensitivity/accuracy of the clock measurements? I seem to remember reading a while back about sideloadable apps that could give more accurate locations, but maybe this was just a hoax?

      The US stopped degrading the accuracy for non-military GPS users in 2000. Surveyors have been able to get better than 10mm accuracy with their GPS units by using a GPS base stations over a known points to provide corrections to their other GPS units.

      No, GPS accuracy is limited by physics, even the military Precision broadcast is limited in accuracy, it's not magic, just tricky relativity.

      The USA reduced the "dither" (artificially introduced "noise") on the civilian frequencies back in the late 1990's (or around there some time) bringing the accuracy for general users down from 100m to about 4m or so (if you gave a clear sight to at least 4 satellites with no multipath artefact).

      The physical wavelength of the radio frequency limits accuracy (even for phase detection systems( for mobile use. Differential gps uses surveyed "known locations" and analysis of the time delay from every satellite which passes through their field of view to determine the error in real time, but even this only gives accuracy to "several" cm, and the local area correction is only available to scientists and subscribers of the commercial services.

      Surveying to "cm accuracy" with gps takes more background work than just turning up and taking a measurement.... sorry for that, bit lengthy.

      Last edited 12/02/16 11:28 pm

        Actually the wavelength of the signal gets you to about 20cm, the problem that reduces accuracy is atmospheric disturbance. If you have a calibrated base station to give you correction factors for the atmospheric disturbance you can get close to the 20cm. This is provided by the CORS network in general but you can set up your own reference station which is what a lot of farmers do for differential GPS.

        The next level of accuracy is to use the carrier wave rather than the signal and this gets you to about 2cm, but this I believe used to be hard to calculate and now they're claiming a much improved algorithm for this. It's clever using the phase of the carrier instead of the data.

          Correct, (I did neglect the specifics intentionally, I used the term "several" to mean an unspecified amount of not too many, because pedants exist, you know how it is...)

          BUT if your clock (mobile or base station) is out by a nanosecond, you get the correct relative position (carrier phase) in the wrong 20cm increment, so you still have physics and clocks to blame for the inaccuracy.....

          Precision agriculture, isn't all that precise (in surveying terms) , and it is usually relative to the base station, not really relative to an earth centric reference frame (in "super high precision")

          UAV's have been at the forefront of fusing IMU data to the GPS signal to work out a position at a much higher temporal accuracy than the GPS can deliver (5 or even 20Hz isn't nearly fine enough for correct positional control it makes the position look the slowest loop in the control system by a long way, 2 orders of magnitude in many systems).

          Many claims of sub decimetre accuracy via GPS are overblown (call it advertising fluff, or misinformed nonspecialists overstating claims of their technological superiority), and if true they are using more than just a simple portable GPS receiver coupled to a base station. (Some of the ground reference points are still surveyed using total stations etc to get a sub-mm accurate relative position.)

          just for laughs.

    GPS has never been very accurate. The article has it right. The american D.O.D mothballed the tech for reducing accuracy years ago. That and it made it accurate to 1km at best. Not 10m.
    Only way to get below 10cm at the moment is Survey Grade GPS and specific techniques.
    So many unknowns in the calculations for position based off GPS.

      So your GPS navigator must give a good result if it results in GPS accuracy to within 1km.

      I think there are a couple of orders of magnitude missing here, call it 50m (worst case, without solar flares) and that makes it useful for streetmap navigation, 100m is next to useless, and 1000m is worse than dead-reckoning. (reference:http://www.gps.gov/systems/gps/performance/accuracy/

    Gee i wonder whether the increasing amount of phone towers has anything to do with it

    couldn't find the research in IEEE Transactions on Control Systems Technology. what's it called ?

Join the discussion!