[OSM-talk] Determining direction of travel and segment connectivity from GPS traces

[Nick Hill]

Lars Aronsson wrote:
> Nick Hill wrote:
> 
> 
>>If we are thinking of creating a system which can determine 
>>valid routes, we will need to determine valid routes for 
>>specific modes of transport.
> 
> 
> Do it the Amazon.com way: People who have been here, traveling at 
> your speed and direction, have later ended up there or there.

The problem I am attacking isn't the problem of generating a suitable 
routing algorithm, but determining turns and what type of trasport a 
segment is suitable for, and the directionality and suitability of a 
segment for different forms of transport.

The possible routes and turning restrictions/directionality of routes 
depend on the mode of travel.

For example, a contra-flow bus or cycle lane. These are common in London.

For example, you can always walk in the opposite direction to car travel 
down a one way urban street. A walking routing algorithm would be insane 
if it would only send you one way down a one way street. Or to send you 
walking along the hard shoulder of an urban motorway (Along the 
direction of vehicle traffic!!).

If we are to extract more inferences from the GPX traces, we will need 
to have a method to determine the mode of transport used in that GPS 
trace then apply that routing information for a given layer.