Hi, do you know how accurate can a position be if I have my receiver in a car while mapping? Can you suggest equipment? I’m guessing a bigger antenna could give better results but these are mainly seen in survey-grade equipment. I’m dreaming of average-case accuracy within 3 meters of actual position. I realize that will likely be somewhat costly.
For now, I have been logging with my Android phone but I cannot trust that for much editing, even if I saw that some streets are likely off. Is the dashboard of my car the best place for the receiver?
Sorry if I should have just used the search function, I have seen the Wiki pages but I think that could be more comprehensive/up-to-date.
My cell phone GPS is far less accurate than my backcountry handheld GPS. But even that one is not as accurate as I’d like to really place paths in the correct position. Even if you have a pretty accurate consumer grade GPS which can achieve 3 to 5 meters accuracy in ideal conditions, terrain and the position of the satellites at any given time can mean less accurate positions in average conditions. Not much we can do about systemic errors caused by terrain and buildings, but we can compensate for lack of accuracy on any one track by averaging many tracks.
For this reason, upload you GPS traces to OpenStreetMap even if they are from a cheap GPS receiver in your smart phone. And when you edit you should be able to see traces uploaded for the area. As the number of traces grow, it becomes possible to more accurately place the highway or path. Not sure how you do that on ID or Potlatch, but JOSM can be set to download and display an overlay of all the GPS tracks in the area you are editing.
If you are trying to very accurate mapping in an area you can spend some time at, a low end purpose built GPS (I have a several year old Garmin eTrex 10) can often be set to do “waypoint averaging”. It appears that if you average a waypoint several times over the course of a day or two you can get to very good accuracy (the device claims about 1m even though I’m not sure its that good). If those waypoints are something visible in the available satellite imagery then you can align the imagery and then be confident that your mapping near that/those waypoint(s) is as accurate as anything in OSM.
In iD, click on the layer icon on the right. The check "OpenStreetMap Traces’.
This gives a “WMS” layer with all uploaded traces. This background layer is also available in JOSM.
See this blog for more information
If you have a nuvi, you can activate an option to record a log in binary format, and then correct the positions using differential GPS. With the post-process eliminating the errors accuracy could be easily sub-meter.
I’m not much experienced with android, but it seems that the problems is that the drivers (ot the hardware) gives you a position, not a signal that you can process (post or inline) to eliminate errors.
Also with the nuvi in binnary loging, you can have both tracks, one pretty to the screen adjusted to the ways, and the other useful to mapping with more accuracy.
Thanks for the replies. muralito, you might be correct, otherwise I would suppose there was some app to do corrections. More hackable phones would sure be nice; well, I think I get the Neo900 when it comes out. It’s hackable and all software is free except for the GSM modem firmware. It has GPS and GLONASS capabilities.
Is any Nuvi just as good as the other? That sounds good! Real-time correction would be optimal of course but if the post-processing is simple enough then why not. Could I batch process a bunch of files to correct the points and export them in a format suitable for Openstreetmap?
I do upload my traces but many places I go to probably won’t have any other people logging data, so I would like to get accurate enough data so I could map the points.
For me, it is not easy to tell what to expect of different equipment. Ideal would be a logger that gives <3 meter accuracy in a car and close to a meter if I step out to log some point of interest, preferably logging with a frequency of more than 1 Hz. I think there probably are some units nowadays that can get real-time accuracy of one meter or less, for a reasonable price, but maybe someone with more experience can chime in here.
Those requirements will be hard to fulfill e.g. in a forest or a narrow street with high buildings. The GPS accuracy depends on the number of visible satellites and their position. This maximum accuracy also depends on your position on the earth.
I understand the limitations inherent… but there is not too much information on how accurate some specific devices are in action. I have read the wiki and looked into different receivers but I don’t get the picture. Could a higher-end consumer GPS be accurate to a meter in an open area, with good weather? What kind of a unit would be accurate to 3 meters or better inside a car when doing mapping while driving. I could shell out some cash, too, as I would have some other uses for the unit as well. I’d like enough precision to be confident enough for moving some roads that are obviously a bit wrong on the map and for more important nodes it would be ideal to get the precision close to a meter if I step out of the car. 10 Hz logging would not hurt either.
I have searched, but I did not find any comprehensive test for GPS units in this range. If anyone has good links, please share
And that is the 50 percentile error not the absolute maximum error, and only for the optimum satellite confiuguration and visibilty.
Although GPS was the enabling technology for crowd sourced mapping, currently, in those places where professional mapping isn’t suitably licensed, Bing tends to be the best source for position and geometry, but even that can have errors that exceed the sort of figures being talked of here.
Beware that some mobile phones will use copyright sources to refine GPS tracks. You must disable any such feature before using the device as a source for OSM. Apple reportedly snaps GPS locations do the nearest road, and Android can use WiFi SSID locations.
