I think this problem has some hidden assumptions about what is the rate
of the incoming points, or more appropriately how close together the GPS
points are and with mis-alignment of GPS with respect to high density
road networks. If the GPS sample rate is high and the relative distance
is small percentage of the road network segments then the problem is
easier to solve because you can assume that any given point is likely on
the same segment as the last point or just transitioned to the next
segment in the network.
There are still serious issue if there is a mis-alignment between the
GPS data and the road segments, especially in dense road network areas.
Here you might need to apply a stickiness to the segment you are one and
do additional analysis and in fact defer judgment to N-1 when looking at
point N.
Also as your sample rate spreads out and the distance between GPS point
increases then the last segment and the current segment might not even
be touching so you might need to route between the adjacent GPS points.
Add to this mis-alignment issues and you have a very tricky problem to
solve.
The above issues are particularly prevalent in CDMA cell phone tower
assisted GPS systems that can have a lot of error in urban areas which
also have higher density road networks, but need to be considered with
any system like this.
-Steve W
On 1/11/2011 12:15 PM, Ralf Suhr wrote:
This will work. The GPS point is never on the road geometry.
ST_Line_Substring( road,
ST_Line_Locate_Point(road, ST_ClosestPoint(gps_start, road)),
ST_Line_Locate_Point(road, ST_ClosestPoint(gps_end, road))
) AS highlight
Gr
Ralf
Am Dienstag 11 Januar 2011, 18:08:29 schrieb Ralf Suhr:
> Hi Martin.
>
> ST_Line_Substring(road, ST_Line_Locate_Point(road, gps_start),
> ST_Line_Locate_Point(road, gps_end)) AS highlight
>
> Gr
> Ralf
>
> Am Dienstag 11 Januar 2011, 17:46:06 schrieb Martin Fafard:
> > Ralf
> >
> > Yes, ok for the buffer. But how to "dynamically segment" my new road
> > from my original segment?
> >
> > Martin
> >
> > Ralf Suhr a écrit :
> > > Hi Martin,
> > >
> > > you will not get a correct solution for this problem. Your road
network
> > > and the gps points differs always.
> > >
> > > A possible "solution" is building a buffer around the streets und
look
> > > für intersection with gps points and filter by direction road/points.
> > > Grouping gps points can give you the start and end point for the
road.
> > > If a road is only one times driven.
> > >
> > > Gr
> > > Ralf
> > >
> > > Am Dienstag 11 Januar 2011, 17:16:18 schrieb Martin Fafard:
> > >> Hi
> > >>
> > >> I have a network of roads (topological) and I receive GPS points. I
> > >> would like to create new lines who follow exactly the same path
as my
> > >> road from the GPS points (to "highlight" where the GPS unit was
on the
> > >> network). The problem is I need to segment the new line because the
> > >> GPS points can stop anywhere on my roads... Any advice to do this?
> > >> Thank you
> > >>
> > >> Martin F
> > >> _______________________________________________
> > >> postgis-users mailing list
> > >> [email protected]
> > >> http://postgis.refractions.net/mailman/listinfo/postgis-users
_______________________________________________
postgis-users mailing list
[email protected]
http://postgis.refractions.net/mailman/listinfo/postgis-users
_______________________________________________
postgis-users mailing list
[email protected]
http://postgis.refractions.net/mailman/listinfo/postgis-users
