Australia will update its latitude and longitude after moving 1.5m
abc.net.au
abc.net.au
I'm sure surveyors, machinists, and physicists would be interested in how "infinitesimal" location data can be obtained. This article doesn't really spell out what is going on. Did Australia use some other datum than WGS84 and now they changed it?
I was going to justify their phrasing as being easier to understand, and accurate enough for a lay audience, but ... mine is still better, IMHO. I would have said:
"Self-driving cars, for example, need more precision than GPS has to avoid accidents."
This is the first stage of a modernization effort in Australia for the datum used there. A lot more information here: http://www.icsm.gov.au/geodesy/modern.html
True, the GPS coordinates might have a errors much greater than this 1.5 metres. But:
(1) bugs can be devious, and I don't trust them to respect this particular insurmountable barrier, and
(2) this error accumulates roughly linearly over time and it is better to make frequent small changes than to wait until you have to make a big, panicked one.
The reason I say this is that the WGS84 frame is maintained by the NGA (US National Geospatial Intelligence Agency) based on a relatively sparse and worldwide GPS base station network, and updated to a new version whenever they deem necessary. (So far we're up to the sixth version of the frame, officially known as "WGS 84 (G1762)" - see http://earth-info.nga.mil/GandG/publications/NGA_STND_0036_1... for the gory details)
Note that "using the (current) WGS84 datum" is a very different thing from "using GPS measurements to define a national datum". At the most basic level, GPS measurements just give you timing information which allow you to work out how far you are from the current set of visible satellites. However, to figure out where you are in some well defined coordinate system, you also need the position of the satellites themselves as a function of time (their /ephemerides/, in the jargon). But wait, there are no absolute physical reference frames, so the positions of the satellites themselves need to be measured with respect to something. This "something" is generally a network of GPS base stations attached to the ground in geophysically stable locations. Depending on which set of ground stations you choose and your processing conventions, you'll compute different ephemerides for the satellites. The NGA have one such set of base stations, and they use it to compute the WGS84 /broadcast ephemerides/ which are broadcast by GPS satellites every 30 seconds. But you can make your own! For example the IGS (International GNSS Service, ugh, acronyms) use a much denser set of base stations all around the world to compute their own ephemerides, giving datum IGS14 (previously IGS08) and I believe there's a public transparent process for nations to contribute to this frame.
So if you want to define a stable and accurate national datum using GPS measurements, I believe the procedure goes something like this:
1) Build a robust set of GPS ground stations, well distributed across the land mass
2) Observe GPS timing signals continuously at all base stations
3) Solve for the positions and velocities of the base stations to realize a terrestrial reference frame defined by your base stations. Some arbitrary but conventional choices must be made here - for example, what's the orientation of the cartesian coordinate axes?
4) Process the GPS observations each day/week/whatever, along with a (potentially less dense) set of measurements contributed by the rest of the world to compute satellite ephemerides.
5) Make the resulting ephemerides publicly available, so people can realize high precision measurements in your datum, by combining with their own GPS measurements.
tl;dr - All measurement is relative. The GPS satellite constellation is used to transfer a coordinate frame defined by some ground control stations (you pick which ones!) back to any other point on the ground.
[edit - formatting]
> 4) Process the GPS observations each day/week/whatever, along with a (potentially less dense) set of measurements contributed by the rest of the world to compute satellite ephemerides. > 5) Make the resulting ephemerides publicly available, so people can realize high precision measurements in your datum, by combining with their own GPS measurements.
Wouldn't it be easier to publish the translation between WGS84 coordinates and your coordinates instead of publishing the whole ephemeris so that the existing broadcast ephemeris could be used and s so that your users wouldn't have to download a new ephemeris as often?
Regardless though, people using high accuracy GNSS equipment are pretty much resigned to needing sideband information to actually get an accurate result. This is to account for various physical effects which can't be easily or entirely removed, such as ionospheric phase delays. It can be from a base station at a known location ("differential GPS" or "RTK"), or from a separate commercially available satellite broadcast eg, Fugro G4 or Trimble centerpoint RTX. The latter two actually compute satellite orbit and clock corrections to the broadcast ephemerides, by processing real time streams from a worldwide base station network in real time!
The ephemerides aren't something you can download every so often, for high accuracy you need something like a few hundred kB a day for the whole constellation, IIRC.
http://www.abc.net.au/news/2016-07-28/aust-latitude-longitud...
Not even DGPS. Really. Don't. A ton of metal moving under its own volition really needs to be looking at the things around it, directly, using some kind of vison. Cameras are good.
Aside from wishy-washy long term tasks like what route to take, or which street has traffic, GPS simply cannot be used.
GPS can fail, be jammed, or lose accuracy. I can't imagine there is any risk of any production vehicles using GPS alone. Certainly GPS matching physical reality will be helpful, but no car is going to drive on a sidewalk and plow through fire hydrants because GPS was off by 1.5 meters.
Before a car can enter a physical space, it should verify with at least two local, independent, physical sensors that the intended space is unoccupied and that no detectable objects are on a trajectory to occupy that space. Our brains do these calculations very easily and with minimal conscious thought. However, our brains' attentiveness and accuracy leave much to be desired.
My fizzy drink fell over. It was a mess. Never forget!
I suppose you get used to it, because now we hardly notice anything below a 4.0 or so.
With GPS having at best three metres accuracy? It's nice that they keep their data up to date, but I'm not so sure self-driving cars have anything to do with it.
Also, from my understanding, you only know the relative position of the two stations accurately, not the absolute world location.
It's obviously desirable not to have a systematic error in how GPS lat/long corresponds to maps but it's within the error of GPS.
With GPS having at best three metres accuracy?
You can get down to 1mm of error with good antennas and long-term averaging-- no good for moving objects. This is how we measure continental drift.
I imagine a guy in a Commodore's outfit climbing a ladder to some bridge (but on the land), making an observation, swatting the dust off a log book with his hanky, making an entry, closing the book, looking around with satisfaction, and then climbing back down the ladder.
I think it's exciting to see examples of continental drift on the human scale. Usually, all the models talk about Pangea and hundreds of millions of years, and our brains have trouble seeing the context.
But I wonder if we don't need a more flexible system, maybe like leap days and leap seconds that get added much more often. Or just have a global reference and an offset for the local datum that can be updated yearly.
IIRC that's the speed South Africa and South America are moving away from each other.
The way they do it that they measure the position of some reference point in WGS84 on every tectonic plate, then they measure the relative position of a bunch of point of interest from the local reference point. At a regular interval, they re-measure the actual coordinates of the reference point in WGS84, and that automatically moves all of its dependent points of interest in WGS84. Tectonic plate movement is why WGS84 is not always the best way to store your points of interest, keeping them in the local reference also helps propagation when an update occurs.
It'll seemingly hit Asia in the guts while part of siberia leaves with Korea moving east. And my poor little city's gonna get squashed and move north to replace some cold Russian city...