Matched percentage: 0 about fmm

[cyang-kth]

@Misavan

There are several problems here.

When I run with your configuration, the program never stops. The problem lies in the configuration file.

In your configuration file, the delta, search radius and gps error you defined is in degree (given your original network and GPS data has unit of degree). You should reduce these values. The delta = 3 equals to 300 km distance in practice. That is why the ubodt file generated is also quite big. The same problem lies with the radius where you need to search for the whole network. You can reduce the delta value to 3km (0.03). GPS error could be 50 meters and the search radius can be 300 meters.

Once you change the configurations, the program will terminate normally. However the original trajectory is still unmatched because

• the algorithm itself cannot handle complicated movements such as move against the direction of an edge, move back and forth
• The GPS trajectory is quite long and the movement is complex. Whenever a single point cannot be matched by the program, the whole trajectory is marked as unmatched.

I have created some figures based on your data to illustrate these observations.

Normal case - part of the trajectory is corrected matched (violet is the raw trajectory and orange is the matched result)

The algorithm has some problem to match in the overpass.

Rather complicated movement