Re: [sumo-dev] Modelling parking

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

Hello,

you are correct in your finding: Parking manoeuvering is not being modelled at the moment.

Note, that the space requirements during roadside parking aren't being modelled either at the moment. There is a fixed number of spots that are usuable regardless of vehicle length.

Showing Manoeuver dynamics visually is somewhat difficult because there are some deep-seated assumptions in the code regarding the driving direction of vehicles (vehicles never drive backwards).

I think a more  abstract approach, that keeps the vehicle on the road for some time after reaching/leaving the parking position and thus accounts for blockage during manoeuvering, should be fairly easy to implement.

To account for driver specific as well as location-specific effects I would introduce a new vType parameter as well as a parkingArea parameter and multiple both values to detemine the manoeuvering time.

E.g.

- parallel roadside parking: 5

- angular roadside parking: 2

- car park: 0 (all maneuvering takes place inside the facility and does not affect other road users)

- experienced driver type: 0.8

- inexperienced driver type: 1.2

Occupancy might be another multiplier (0.5 + relativeOccupancy) but I suppose you will want to calibrate a more complicated function if you can get the empirical data.

For the implementation you should look at

MSVehicle::processNextStop (determines whether a stop has been reached and how long it should continue)

MSLane::executeMovements (Line 1741 transfers a vehicle from the road to a parkingArea)

MSVehicleTransfer::checkInsertions (transfers a parking vehicle back onto the road)

regards,

Jakob

Am Mo., 17. Juni 2019 um 00:10 Uhr schrieb The div <div@xxxxxxxx>:

Hi,

I've a personal interest in parking behaviour and have been looking at SUMO with a view to modelling this.

I'm very new to the code and my C++ is seriously rusty, so my apologies if I've missed the blindingly obvious in what feels to be a solid, very comprehensive solution.

Stepping through the code it appears that parking in the current model does not include manoeuvering time.

My perception is that in the simulation step, after the vehicle stops, it is effectively moved into the parking bay and removed from the net, irrespective of on-road/off-road, space angle and forward/reverse/parallel parking factors.

If I have misunderstood this please could someone give me a hint as to where in the code I might look for more granular behaviour?

If this isn't modelled then I'd like to spend some time learning SUMO/investigating this and would be grateful for any observations/pointers. I'm specifically interested in impacts of bay angles and in-lane bays.

( I suspect that lane disruption related to eg parallel parking will impact congestion/timings and consequently emissions, similarly reverse parking is assumed to reduce emissions because the manoeuvre is performed with a 'warm' engine.)

I might think that there would need to be:

Driver parameters:  encompassing experience/skill and/or (cultural?) propensity to reverse park.

Vehicle type parameters:  for nominal time to forward/reverse/parallel park, and equivalent timings for leaving the space

The manouevre time derived from the current bay occupancy and these parameters would then be used to derive the lane blocking time associated with beginning / ending parking.

My initial guess would be that this could be implemented in the vehicle class but I don't yet understand the code enough to know whether this would be the 'correct' place - or whether it may be more appropriate eg in the lane change classes.

cheers (& thanks for reading)

div

_______________________________________________
sumo-dev mailing list
sumo-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/sumo-dev