Explore higher spatiotemporal realism in DRT bookings

[nkuehnel]

Currently, agents in a stop-based DRT service already decide where to walk to during replanning. During QSim, agents will request for immediate rides upon arrival at the stop.

In reality, agents would typically request their ride before their access walk during the previous activity. Many providers will only then decide where the agent will be picked-up. This dynamic stop selection is currently not possible in MATSim as the stop is pre-selected and fixed (a quite simple approach could be to test the two or three closest stops for insertion and select the best out of these).

This would require some larger changes to workings of DRT as the access and egress walk legs would essentially become within day decisions for agents choosing DRT.

The temporal component, for agents being able to request rides a few minutes later in the future is currently addressed by Tarek and Sebastian, I believe (it's essentially prebooking).

Happy to discuss the feasibility of this beforehand.