[OSM-dev] Feasibility of an 'in the field' mobile editing app?

[Igor Brejc]

Nick Whitelegg wrote:
>> Sounds nice, but there's one problem: creating OSM ways automatically 
>>     
> >from GPS data (without user editing) is not a good idea, for several 
>   
>> reasons:
>> - GPS (in)precission: it is a good practice to cover certain path 
>> several times before actually creating a way
>>     
>
> Do many of us actually do that though? I have to admit I don't, as it 
> would simply take more time than I have available. In the countryside in 
> particular (which is what I was thinking of), one needn't be that bothered 
> about pixel-precision accuracy anyway. And in the town where I live, I 
> find the streets I've created generally match up well with the Yahoo! 
> imagery.
>
>   
Well we should strive towards those ideals as much as we can. From my 
experience with my Garmin unit, sometimes the GPS fix is bad (>20 m), 
and this kind of inaccuracy then makes difficult the stuff you were 
suggesting (like server-side checking for duplicate ways). What I miss 
in my unit is recording of precision in GPX files.
>> - Depending on your GPS tracking settings, you could end up with a lot 
>> of unnecessary OSM nodes (while standing still, for example). This could 
>> be fixed by some data cleaning SW, I suppose, but it should be taken 
>> into account.
>>     
>
> The idea was to use some sort of path simplification algorithm to avoid 
> this happening.
>
>   
Agreed, but this should be best implemented on the client side, to avoid 
excessive uploading of unneeded data.

-- 
http://igorbrejc.net