Re: [sumo-user] Potential danger in not using minGap

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

It depends on the car-follow model.

The Krauss model treats the minGap as inviolable, hence setting it to 0 does not lead to crashing as long as tau > action-step-length. The downside would be that the cars are extremely close when in a standing queue (because the tau value only affects moving vehicles).

Other models such as IDM do not guarantee minGap so you might see crashes at low speed (just a guess, I did not test this).

A recent analysis by my colleague of jams in the highD dataset (https://www.highd-dataset.com/) gave a fit of

gap = 3.9m + 1.04*speed (with high variance, though)

regards,

Jakob

Am Di., 29. Sept. 2020 um 01:56 Uhr schrieb Bae, Jong In <jbae42@xxxxxxxxxx>:

Hello,

Would there be any potential danger in not using minGap (set it to 0) and increasing tau to like 1.8s or 2s instead? 

I assume that using the tau value greater than 1 should also give the follower vehicle enough room to brake without crashing.

Basically, I'm curious if larger tau values can cover what minGap is supposed to do.

What are some factors that I need to consider before implementing such changes?

Should I expect some unrealistic (or undesirable) behaviors after this change and why?

Thank you so much in advance!

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user