Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×
Build Technology

Centimeter-Resolution GPS For Smartphones, VR, Drones 63

agent elevator writes: UT Austin engineers have come up with a software fix that corrects for the errors GPS has when using the tiny antennas on smartphones. They demoed it using a VR setup and got 2-cm accuracy. For now it runs on a separate processor from the smartphone, but they say they'll fix that. The demo appears to have been done on a rooftop. VR. Outside. On a roof. Doesn't seem like a good idea, does it?
This discussion has been archived. No new comments can be posted.

Centimeter-Resolution GPS For Smartphones, VR, Drones

Comments Filter:
  • by beschra ( 1424727 ) on Thursday May 07, 2015 @03:26PM (#49641493)

    Put a question in the summary that only makes sense if you RTFA. Nice.

  • > They were also able to precisely track a virtual reality headset with the same precision.

    One does not "precisely track" a VR headset with two centimeter resolution. I'll guess that they continued to use the IMU tracking that is built into the Samsung Gear VR, and they used it to display the tracking of external objects that were measured with two centimeter resolution.

  • by 140Mandak262Jamuna ( 970587 ) on Thursday May 07, 2015 @03:29PM (#49641515) Journal
    The cell phone GPS antennae are tiny they catch the signal from the satellite and also many reflections. These reflections confuse the processor trying to fix the distance between itself and the satellite. They seem to have developed some signal processing algorithm that would remove these reflections. The article is skimpy on details.
    • by Gen-GNU ( 36980 ) on Thursday May 07, 2015 @03:37PM (#49641579)

      The link in the article, to here [gpsworld.com], gives a much better description of the SP algorithm. In fact, it's a much more informative article, but it doesn't have a picture of a guy with a cell phone strapped to his face.

    • by Anonymous Coward on Thursday May 07, 2015 @04:36PM (#49642075)

      The key thing is that cm level positioning is only possible using differential techniques. You need a receiver at a fixed position and you then correlate the carrier phase measured by your mobile receiver to your fixed receiver. This is a technique known as carrier-phase differential positioning.

      The problem is that carrier phase techniques are extremely sensitive to signal degradation - particularly multi-path signals (reflections from the ground/buildings/etc.) The conventional "code phase" technique is, in contrast, reasonably resistant to such signal degradation.

      The problem then with carrier-phase techniques is that you need a good antenna. Traditionally, this technique has been restricted to survey grade equipment, in order to be able to track the phases accurately and resolve any ambiguities in the phase measurements (which can be huge in a multipath situation). This becomes much harder with lower quality patch antennas, and essentially impossible with a smartphone grade antenna (10 minute ambiguity resolution times using conventional techinques).

      The paper described a combination of 2 techniques: 1. attempting to resolve phase ambiguities for each individual pair of carriers (satellite signals), something previously reported and 2. adding small random wavelength scale motion to the antenna, which can then be correlated to the carrier phase residuals in order to speed ambiguity resolution. The authors present data showing that by using these techniques, the phase differential ambiguity can be resolved, even with a smartphone antenna, within 20 seconds.

    • Not that complicated. The multipath signals always lag the direct path, so the DSSS correlation distinguishes them. First one in from each sat is the correct one.
    • From my understanding no cellphones that I am aware of have real GPS, but rather GPS "Assist".

      Which means that the cellphone doesn't really connect with any satellites whatsoever. They communicate with cell towers which are located at known GPS locations. GPS Assist is able to determine your location by communication with these towers at known locations to work out where it is. About the only thing that would improve that (apart from having real GPS), would be to improve the number of cell towers and maybe

      • Most smartphones do have "real" GPS receivers in addition to Assisted GPS [wikipedia.org]. How long it takes to get an initial GPS fix depends, in part, on how well the device can predict your location, as well as up-to-date knowledge of the satellite orbits; A-GPS takes advantage of cell tower data to provide an approximate starting point for the GPS and a faster way to download the orbital information and thus get a quicker fix. A GPS receiver is still necessary for a precise location, and my Nexus 5, to pick one example

  • One of the main error sources for GPS is the propagation delay of signals that go through the atmosphere. An antenna can be in the exact same location but report a different location from day to day. This error is usually eliminated using http://en.wikipedia.org/wiki/Differential_GPS">Differential GPS [wikipedia.org].

    Differential Global Positioning System (DGPS) is an enhancement to Global Positioning System that provides improved location accuracy, from the 15-meter nominal GPS accuracy to about 10 cm in case of the best implementations.

    DGPS uses a network of fixed, ground-based reference stations to broadcast the difference between the positions indicated by the GPS (satellite) systems and the known fixed positions. These stations broadcast the difference between the measured satellite pseudoranges and actual (internally computed) pseudoranges, and receiver stations may correct their pseudoranges by the same amount. The digital correction signal is typically broadcast locally over ground-based transmitters of shorter range.

    I see no mention of using differential GPS in this system.

    • by Anonymous Coward

      I understand that even the military GPS receivers use DGPS to achieve centimetre accuracy. Professional survey grade instruments are good to about 2m without a base station, so it is unlikely a phone is going to better that by 2 orders of magnitude. Possibly the article is meaning precision rather then accuracy (comparison of one position to the next, not to the real world).

    • This is likely a RTK system which is a bit more advanced than DGPS. A lot of the hardware necessary for DGPS can also be used for RTK as both require a base/reference station. The difference is that DGPS assumes that the correction of the known reference station is also applicable to the roving station, RTK makes use of the L2 carrier phase data from the reference station and roving unit and will use that to correct for the atmospheric differences. In both cases you need to get data from the reference stati
      • In both cases you need to get data from the reference station in real time

        I didn't see them mention a reference station at all.

        • Well if it is on a smart phone having a data connection would allow the retrieval of the necessary data so it may be that who ever wrote the article isn't really aware of what is needed. Yes the antennas in phones suck, as do the GPS chipsets. Also whoever wrote the article doesn't' seem to know much about antennas as I use some very small but pretty good [synergy-gps.com] ones with nice uBlox LEA-6t [ning.com] module in my home built setup for RTK and they work great.

          I wonder, after carefully reading the article, if they are discuss
  • by Kaenneth ( 82978 ) on Thursday May 07, 2015 @03:47PM (#49641651) Homepage Journal

    Well, as long as you don't get closer than 2 centimeters from the edge.

    That would be a good test for a GPS product; have someone navigate a dangerous area blindfolded using it's directions.

    • Re: (Score:3, Funny)

      by Anonymous Coward

      (Steps over the edge)

      "Please make the next available safe U-turn."

  • Such a demo makes sense because you compare the performance against a control (without the software fix). Real-life improvement and absolute performance are simply a different, farther-reaching question.

    The other important thing to note here is how well this could perform in combination with the wide-area augmentation system (WAAS), which corrects for atmospheric variation of signal travel time by using information the is generated by ground stations and broadcast by the satellites. WAAS isn't included

    • by Anonymous Coward

      This is a differential system. It consists of a mobile receiver, and a fixed receiver close by, which is at a precisely known position. The signals from the mobile and reference receiver are then compared and corrections can be applied.

      A crude differential system can get resolution down to about 1 meter. Space based augmentation systems (for example WAAS) are unable to transmit precise location specific correction information, so only track gross large scale errors.

      For cm precision, you need to use a differ

      • It *IS* a cell phone, so there are normally at least 3 nearby fixed reference stations in local communication with the handset, no? All of those provide GPS-disciplined time signals. The only question is whether they actually provide reference offsets. They certainly could, if the software were present.
        • I believe they're doing that already (A-GPS) and that's used by cellphones (sometimes triangulation of towers works crudely when GPS doesn't). But that's an augmentation system not unlike WAAS, and what the anonymous expert above explained is very different (A-GPS can't deal with signal reflections off objects, and that's logical).
        • Actually, where I live (Phoenix area) it is not uncommon to find only one fixed reference station (cell tower) connected to my phone.

          T-Mobile begs me to turn on WiFi scanning to permit improved accuracy. WiFi is a nuisance on my phone, as I'm around at least 3 networks daily that all ask me to log on, and if I don't I miss email, texts, and updates. A nasty little feature of how Android handles secured WiFi.

      • Super interesting. Thank you.
  • Nice... (Score:5, Informative)

    by Anonymous Coward on Thursday May 07, 2015 @04:40PM (#49642125)

    if you are on a rooftop.

    I have made my diploma in geodesy and if it would be so easy to get 2 cm with smartphone everywhere then it would be rolled out already. What they are showing is known and realized for like 6 years already, so nothing new. The positions problems will still persist on street level where like 99% of all peoples move and cars drive? The problem is not the antenna its the signal interruptions and multi-path reflections you have on the streets that you cannot simply fix with software as the amount of unique signals is limited.

    It will actually be fixed by more satellites integrated into one software equalization equation and different signal modulation.
    The first is taken care of by the GPS race between all superpowers entangled in it. By 2018 we will have around ~100 GP(S) Satellites in stationary or modular orbits. The only limiting factor here is the software of vendors which has to align and process the public signals GLONASS (RU), GALILEO (EU), GPS (US), BEIDU (CHINA), Indian and Japan signals. Which no one does atm, it is done at university level (for instance german TU Dresden).

    The second one could improve position but just if all countries involved in GPS development and evolution would work together but actually they are not. Instead they compete for the frequencies ...very retarded game but that's another story.

    What people need to understand is that most position software out there barely manage to get a better position out of GPS + GLONASS. There is nearly non that handles the four major systems because even these systems have, based on different earth models, different results for the same position.

  • The actual paper (Score:4, Informative)

    by Anonymous Coward on Thursday May 07, 2015 @04:52PM (#49642229)

    https://radionavlab.ae.utexas.edu/images/stories/files/papers/ion2014Pesyna.pdf

  • by Catbeller ( 118204 ) on Thursday May 07, 2015 @05:43PM (#49642625) Homepage

    We can be tracked with decimeter precision. Yay. I'm sure it won't be used against us. Carry on.

  • That would bring immediate improvements to GPS accuracy, which was initially intended as a remote terrain guidance system for nuclear tipped cruise missiles and submarine launched ICBMs.

    (Yes, I know that then-President Clinton issued a directive to abolish Selective Availability in 2000, but that doesn't go anywhere near explaining why the average handheld still can't get better than twelve feet horizontal accuracy when a cruise missile can use GPS - travelling at 500mph - to pilot itself in through a parti

    • What do you mean? The L1 signal is no longer randomized but the L2 is still encrypted for military use. However the L2 phase data can still be used it just takes more processing to align it but if the data wasn't encrypted it would be trivial to align which is what military GPS units do.
  • by chasm22 ( 2713399 ) on Thursday May 07, 2015 @07:24PM (#49643301)

    They use RTK technology. It doesn't rely on the same data as GPS or dGPS. It isn't new techology. And it wouldn't matter if the DOD screwed with the dGPS signal.

    What is new is just what the article talked about. They have found a way, through 6 years of developing the software, to use a cheap antennae to capture signals. Prior to the software development, cheap antennae couldn't be used because they allowed for too much signal degradation. Their solution is their software, which they have apparently perfected to the point where it can recognize and correct or mistakes caused by the antennae.

    My understanding of RTK is that it isn't useful for navigation, but is very useful and accurate in obtaining your position(while not moving) This is because it relies on a base station in addition to the other receiver.

    • Well not entirely. It can be used for navigation but it requires actively communicating with a base station or set of base stations and getting their raw pseudo range data as well as the L2 carrier phase data. From there with some fancy processing and enough compute power you can basically get real-time position accuracy. There is a set of open source programs and libraries [rtklib.com] that can provide the software necessary. Also if you don't want to setup your own base station there may be one nearby or a set that ca
  • by Anonymous Coward

    UT Austin engineers have come up with a software fix that corrects for the errors GPS has when using the tiny antennas on smartphones. They demoed it using a VR setup and got 2-cm accuracy.

    Confused multiple test results there, didn't ya? The 2-cm results were achieved when they used a separate GPS antenna (not a smartphone GPS) stuck on top of the Samsung Gear in all the photos. The quality of antenna in that little black box is at least an order of magnitude better than those in typical smartphones.

  • by lordlod ( 458156 ) on Friday May 08, 2015 @03:10AM (#49645015)

    Part of the problem that isn't addressed in the summary is that to have a cm accurate position you also need to have an oscillator that is accurate in the tens of picoseconds range.

    From the article:
    > The clock attached to the external front-end was an oven-controlled crystal oscillator (OCXO), which has much greater stability than the low-cost oscillators used to drive GNSS signal sampling within smartphones.

    An OCXO is far more expensive than a smartphone manufacturer will happily absorb (~$30). It is also constantly heating the crystal so your battery life gets thrown out the window too.

    GPS manufacturers very carefully select their cheaper TCXO chips in order to get nanosecond accuracy. Special tricks are used to get sufficient DAC resolution on the voltage control in order to steer them to the correct level. I have been out of the industry for several years but I would be shocked if there has been a 100x improvement in quality without hearing anything about it.

When we write programs that "learn", it turns out we do and they don't.

Working...