Sampling point router¶
This router is responsible for finding the best route between two given sampling points \(Start\) and \(Goal\).
It is composed of the following three operations.
1. Possible routes calculation¶
Because a sampling point may have more than one candidate and a driver can cross the projected point in two directions, there are many possible options how to begin the routing between the two sampling points, as we see in Fig. 20.
Fig. 20 Motivation¶
The underlying Directed candidate router needs a single sampling points selection which consists of the following data:
The two sampling point candidates \(Start\) and \(Goal\).
Direction information for each sampling point candidate.
The set of possible sampling points selections is reduced by two givens (see also Routing):
The initial direction of the \(Start\) sampling point, induced by the preceding route (if it’s not the first sampling point of a new route).
The allowed travel directions.
Example¶
In the following example, the current route ends at sampling point candidate \(C_2^{Start}\), that’s why routes from other candidates (here \(C_1^{Start}\)) are not considered:
Fig. 21 Overview¶
In Fig. 22 and Fig. 23 two possible routes have been calculated by the underlying router for two possible directions of \(C_2^{Goal}\).
Fig. 22 Possible route 1¶
Fig. 23 Possible route 2¶
If Router filter’s allowSelfIntersection is set to true, also the following route would be possible:
Fig. 24 Possible route 3¶
2. Routes clustering¶
All found routes are now clustered as described in Clustering.
3. Route selection¶
The role model of all clusters are then sorted as described in Final Evaluation determining the final, most realistic route.