[SoC] My 4th week report

Stephen Woodbridge woodbri at swoodbridge.com
Wed Jun 16 10:05:13 EDT 2010


Ashraf Hossain wrote:
> Hi All,
> 
> This whole week I studied only the standard data format which will be 
> easily incorporateable to the engine and also reduce the complexity.
> I got that if we use a base street layer and upon that we use different 
> layer for different public route and if the both layer share the common 
> nodes(end points of a feature are treated as node in graph), then the 
> execution complexity is reduced.
> 
> But I noticed that the public transit like train station and river port 
> are generally far away from the street. At that case this node will be 
> unreachable or we have to apply a search of closest node which will make 
> this more complex. But instead of this if we add simple walking layer 
> which connects the nodes from street to station and port, then we can 
> generate a very smart path with lesser time.
> 
> I also noticed that the bus route if we treat the whole polylines 
> between two bus stops as single feature and the end points are end 
> points of the general street layer then the execution complexity will be 
> reduced as I stated before. 
> 
> Next week Task:
> -------------------------
> I am implementing the public route support with my proposed data format, 
> next week without time support. I will use simple dijkstra's algorithm 
> for this.
> 
> Problem:
> --------------
> My only problem is that I do not have the standard data format I 
> proposed but I will make this for myself.
> 
> 
> My Details report is as follows.
> 
> http://sourceforge.net/apps/trac/opengraphrouter/wiki/GSoC2010_report_20100614

Ashraf,

I think this is a good analysis of the problem. I think you should 
assume that the design of figure one is the way to go and that data 
provides will be able to supply data in this format or have it edited to 
work in this format. Also you can assume that eventually some of these 
edits and data fix-ups can be done in the graph building process in the 
future. For example splitting street segments to add a node at a bus 
stop is a good one that could be done in the data loaded with some 
additional analysis and assumption.

You say that each bus route should be a single polyline. I think that 
each segment between stops should be a single polyline and hence a 
single segment in the graph.

By the way, I love the graphics, they are very professional looking.

Good job,
   -Steve


More information about the SoC mailing list