GPS positions to route mapping

[RalfGelfand]

Hi,
I???m working for a software company and we are designing a new software solution for one of our customers. In this solution road maps and routing will take a serious part.

Based on the ArcGIS API documentation I am quite sure most of our map related requirements can be handled with ArcGIS. Some details are still not so clear to me yet.

Now the question:
A centralized system periodically receives positioning data from moving cars. It needs to track the car???s actual route. As you know GPS data is not very accurate. Just finding the route a car took by looking at each single positioning data separately and connecting these measures points to a route, might have funny resulting effects. These resulting routes will be incorrect especially in areas where roads are close together or are overlapping (bridges, tunnels, etc.). The route information so far and the road map have to be taken into account to figure out the most likely route a car took.
Even if the transmitted position information suggests a car is currently on the road underneath a bridge, the car would most likely not have jumped of the bridge if the previous position was on it. The time difference between the measures makes it impossible to find a valid route from position on the bridge to the suggested new position underneath the bridge. This means the measure of position underneath the bridge is not accurate and has to be corrected to be on the bridge again.

Does ArcGIS provide such functionality to smoothen GPS data and to build realistic routes? Or alternatively what high level functionality is provided to support building such functionality? Our central system would have to do this for hundreds of cars and would have to do some further calculations based on the resulting routes.

Thanks,
Ralf

[MarkBaird]

If you have really bad GPS signals which jump all over the place in an area which roads are very close together, then putting these points through a routing engine might result in a route which goes all over the place.

However are your GPS tracks really this bad?

One thing you could try it taking your jumping points and turning them into a Polyline geometry.  What you can then do is use the GeometryEngine class to perform a simplify operation which would smooth out any blips.

The GeometryEngine also allows you to find the nearest point along a nearby line.  Take a look at the Java Doc:

https://developers.arcgis.com/en/java/api-reference/

Does this help?

Mark

[RalfGelfand]

Mark,
thanks for the answer. The GPS signal can easily drift up to 20 meters. This is enough to potentially result in strange routes. I will receive a signal every 20-30 seconds and have to build a route as accurate as possible. I think to smoothen the points with geometric methods will help but might not result in the best result possible.
My main objective with this question was to see if ArcGIS has a build in functionality for that kind of problem or if it provides methods greatly supporting us in building such functionality.
Based on you answer and the absence of other answers bring me to the conclusion that I did not simply miss the related section in the documentation. All in all this is fine with me and I simply have to be aware of that when doing the project estimation.

Thanks,
Ralf

[GayleYoung]

Hi,
you could consider having your Network Dataset model z-elevation or z-level values; that would mean routing will automatically distinguish bridges from the roads under them. Also, as you have probably found, you can define an impedance for each route analysis (such as quickest, shortest, most use of major roads through useHierarchy), if you have the supporting data available.
Gayle