It depends on the requirements. I think mapping from Bing imagery is quite reasonable and that isn't done to 1 cm accuracy.
However as you point out there are other solutions and I'm not advocating one specific solution merely that it exists. If there is a requirement that needs high accuracy then this might work. Perhaps it needs a wiki entry to list the advantages and disadvantages of different approaches. Certainly locally imagery is much better than GPS near tall buildings. Another issue is if you have two things in the map one that has been mapped with high accuracy and one not there isn't really anyway to tell which is the high accuracy one. Cheerio John On Sun, May 5, 2024, 14:30 Greg Troxel <g...@lexort.com> wrote: > John Whelan <jwhelan0...@gmail.com> writes: > > > Search Youtube for "Andreas Spiess ESP32 precision GPS receiver > > (incl. RTK-GPS Tutorial)" I deliberately haven't put a direct link > > in. > > > > It needs packaging and documenting but I think this sort of > > differential GPS could be very useful in accurate mapping. > > comments without watching :-) > > RTK can be done for not a huge amount of money, less than a moderate > phone and a lot less than a new high-end phone. The u-blox F9P is > about $200. A decent antenna is another $100. You need power and > cabling of course. You then need to get RTK reference data and inject > it, and the F9P will compute RTK solutions. I have an F9P; with good > reference data and a decent antenna, it works very well with clear sky > and ok in the forest, getting you fairly reliably within about a 1 m > vs 10 m non-RTK. > > If you don't have a local RTK network reference you can run your own, > with double the hardware and some work to find the location of your > fixed antenna. > > 1 cm is a bogus claim; I have returned to the same piece of rebar > multiple times and the 30-sec averages cluster in a diameter of about > 4 cm. I do not know if there is systematic offset. People tend to > believe the accuracy numbers they get out of the device and believing > them does not make sense. > > One can also use rtklib with a receiver that outputs raw, without > needing an F9P. This is less expensive and a lot harder. > > This item it out of stock, and a fairly big markup, but > - it works (I have a report from a reliable nerd who has mapped > with it!) > - it has a battery, a case, a display, a uSD for logging > - it has an ESP32 to manage the F9P > - sparkfun has been GREAT about actually working on the firmware > https://www.sparkfun.com/products/18442 > https://github.com/sparkfun/SparkFun_RTK_Firmware/ > > cm accuracy is not really useful for mapping vs 5 cm or 10 cm, but > having 10 cm accuracy means you can just take the data as accurate. > trails that you thought you mapped well turn out not to be where the > map says. mapping done with RTK is so much neater and cleaner. It's > reasonable to map individual rocks and then the map looks like the > ground and lines up with the imagery. If you haven't tried it you > don't know what you are missing. I don't like to map trails with > regular GPS any more. > >
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
