I kinda wonder if the path created by the inertial system is being used to correct the sequence of GPS derived points. Reason being the inertial systems error is tiny, but cumulative over time. Where the GPS error is large per measurement but tiny when averaged over a set of geographically spaced points.
You have two errors, plain noise that causes the position at a particular location to vary. And a location/time[1] specific error that is an offset. If you know the shape of your path precisely then you can rectify both.
[1] Location due to interference of things like trees, buildings, etc. Time because the positions of the satellites changes over time.
You can get a millimeter accuracy without DGPS, but you need several hours worth of signal for a single spot.
This is not a new tech (or a math rather), it's been around since the early 90s if not longer, and it has interesting practical applications like measuring dam movements.
It was in mid 90s, but I can guarantee you that I sat next to a math PhD guy who worked exactly on what I described, including the dam deployment. Single receiver, 3mm precison from 3-4 hours of data, no DGPS. This wasn't in the US though.
Very interesting data. It was cool to notice some graphs had much higher variability and then notice that they were all sites that are far from the equator. Then the article helpfully explained precisely why they had more variance!
Please, can somebody explain these integers that are being resolved? The paper is behind a paywall and the article, of course, offers no clarification. I am almost sure the author of the article has not read the paper for themselves.
They are referring to integer ambiguity resolution, which is the final step in getting a dual-frequency differential GPS position down to centimeter accuracy.
This [1] is a good explanation. It does require a GPS receiver capable of carrier phase tracking, which your cell phone isn't capable of. The cheapest I've seen is this uBlox [2].
Skimming through the article this looks like better signal processing coupled with the usual sensor
fusion (from accelerometers and other sensors).
My knowledge of location systems only extends so far, but from what I know 3G, 4G and WiFi have not
introduced anything particularly remarkable here.
Wifi APs and various BSs are used as beacons (a la Skyhook), but their role doesn't extend
much beyond that.
Anyway, there is still hope that 5G or some IEEE standard will make a difference here.
They could perhaps provide something akin to a ground satellites, maybe at a very low frequency (to
get past obstacles), or/and perhaps using UWB (helps with multipath, TOA determination).
And then maybe we would have zeroish-time-to-fix, indoor-location on cheap, low-power devices.
Do you have any references for the ground-satellite abilities for 5G? I couldn't turn up anything on Google.
Locata [1] has been in the high-end space for a while. I had heard of a lower-end approach too, though the name escapes me, and they were only getting ~50cm.
Edit: The other one I was thinking of was decaWave [2], they claim 20cm in 2 dimensions.
I would like to see some location improvements in 5G, but I've no idea on whether or not they will
be there. (The little I've heard on 5G has been from non-technical types, always as a buzzword).
From my limited knowledge it would nice if they provided a signal somewhat similar to GPS.
Perhaps at a low frequency, as to avoid reflections and attenuation (or just a UWB signal).
The main advantage of having this together with 5G would be deployment, we would piggyback a location system
on top of 5G.
Back to your comment, I did know about DecaWave, but not Locata.
Locata looks interesting, it really does seem like GPS brought to earth.
One good thing about GPS type systems is how well they scale. The satellites just provide
the signal, they don't even have to know how many receivers are out there. (Also good from a privacy standpoint.)
DecaWave system is loosely based on IEEE 802.15.4a, which uses UWB. I think it
does two-way ranging (basically a very precise ping). And since they use UWB, their
system seems to be quite resilient to multipath.
I think this is a little misleading. The IEEE paper is about a reducing the computational effort for a filtering technique (Contemplative Real-Time) developed by one of the authors, which seems to be published here [1].
The original paper in compares the CRT approach to a more traditional Extended Kalman Filter (EKF) approach. The EKF is very efficient but it does trade of some accuracy, because it requires a linear approximation of the system to be used. Both the CRT and EKF are used to blend or fuse inertial measurement data with GPS.
To achieve the claimed centimeter-level accuracy requires a local (within ~20 miles) base station and a data connection to the base station to achieve the levels of accuracy being discussed. Depending on the quality of the GPS receiver there will be a few seconds up to 10s of minutes convergence period for the GPS filters to initialise.
The results presented in [1] show that using a moderately expensive (dual frequency, code tracking) GPS can approach (not exceed) the accuracy of a high-end GPS (dual frequency, code and carrier phase).
In summary, this is a interesting incremental improvement, not a huge breakthrough that's going to bring centimeter accuracy to cell-phone GPS tracking.
"Farrell said these requirements can be achieved by combining GPS measurements with data from an inertial measurement unit (IMU) through an internal navigation system (INS). In the combined system, the GPS provides data to achieve high accuracy, while the IMU provides data to achieve high sample rates and high bandwidth continuously."
I'd call that centimeter accurate positioning, but still metre resolution GPS.
24 comments
[ 4.6 ms ] story [ 46.5 ms ] threadI kinda wonder if the path created by the inertial system is being used to correct the sequence of GPS derived points. Reason being the inertial systems error is tiny, but cumulative over time. Where the GPS error is large per measurement but tiny when averaged over a set of geographically spaced points.
You have two errors, plain noise that causes the position at a particular location to vary. And a location/time[1] specific error that is an offset. If you know the shape of your path precisely then you can rectify both.
[1] Location due to interference of things like trees, buildings, etc. Time because the positions of the satellites changes over time.
I assume you couldn't get centimeter accuracy without DGPS?
This is not a new tech (or a math rather), it's been around since the early 90s if not longer, and it has interesting practical applications like measuring dam movements.
Without a local reference station, atmospheric and ionospheric disturbances would totally overwhelm the effect they are trying to measure.
This [1] covers it quite well.
[1] http://pasadena.wr.usgs.gov/office/hudnut/SRL/
It was in mid 90s, but I can guarantee you that I sat next to a math PhD guy who worked exactly on what I described, including the dam deployment. Single receiver, 3mm precison from 3-4 hours of data, no DGPS. This wasn't in the US though.
[0] http://www.gps.gov/systems/gps/modernization/sa/
This [1] is a good explanation. It does require a GPS receiver capable of carrier phase tracking, which your cell phone isn't capable of. The cheapest I've seen is this uBlox [2].
[1] http://www.navipedia.net/index.php/Carrier_Phase_Ambiguity_F...
[2] https://shop-america.u-blox.com/en/usd/3~130~AMERICAS/Positi...
https://www.reddit.com/r/technology/comments/45jd4r/scientis...
My knowledge of location systems only extends so far, but from what I know 3G, 4G and WiFi have not introduced anything particularly remarkable here. Wifi APs and various BSs are used as beacons (a la Skyhook), but their role doesn't extend much beyond that.
Anyway, there is still hope that 5G or some IEEE standard will make a difference here. They could perhaps provide something akin to a ground satellites, maybe at a very low frequency (to get past obstacles), or/and perhaps using UWB (helps with multipath, TOA determination). And then maybe we would have zeroish-time-to-fix, indoor-location on cheap, low-power devices.
Locata [1] has been in the high-end space for a while. I had heard of a lower-end approach too, though the name escapes me, and they were only getting ~50cm.
Edit: The other one I was thinking of was decaWave [2], they claim 20cm in 2 dimensions.
[1] http://www.locata.com/technology/locata-tech-explained/
[2] http://www.decawave.com/technology
From my limited knowledge it would nice if they provided a signal somewhat similar to GPS. Perhaps at a low frequency, as to avoid reflections and attenuation (or just a UWB signal). The main advantage of having this together with 5G would be deployment, we would piggyback a location system on top of 5G.
Back to your comment, I did know about DecaWave, but not Locata.
Locata looks interesting, it really does seem like GPS brought to earth. One good thing about GPS type systems is how well they scale. The satellites just provide the signal, they don't even have to know how many receivers are out there. (Also good from a privacy standpoint.)
DecaWave system is loosely based on IEEE 802.15.4a, which uses UWB. I think it does two-way ranging (basically a very precise ping). And since they use UWB, their system seems to be quite resilient to multipath.
The original paper in compares the CRT approach to a more traditional Extended Kalman Filter (EKF) approach. The EKF is very efficient but it does trade of some accuracy, because it requires a linear approximation of the system to be used. Both the CRT and EKF are used to blend or fuse inertial measurement data with GPS.
To achieve the claimed centimeter-level accuracy requires a local (within ~20 miles) base station and a data connection to the base station to achieve the levels of accuracy being discussed. Depending on the quality of the GPS receiver there will be a few seconds up to 10s of minutes convergence period for the GPS filters to initialise.
The results presented in [1] show that using a moderately expensive (dual frequency, code tracking) GPS can approach (not exceed) the accuracy of a high-end GPS (dual frequency, code and carrier phase).
In summary, this is a interesting incremental improvement, not a huge breakthrough that's going to bring centimeter accuracy to cell-phone GPS tracking.
[1] https://scholar.google.com/citations?view_op=view_citation&h...
"Farrell said these requirements can be achieved by combining GPS measurements with data from an inertial measurement unit (IMU) through an internal navigation system (INS). In the combined system, the GPS provides data to achieve high accuracy, while the IMU provides data to achieve high sample rates and high bandwidth continuously."
I'd call that centimeter accurate positioning, but still metre resolution GPS.
Either way, it's a potentially useful advance.