Thanks Jakob, I'm going to try that and will get back to you. Regards, Quentin
2014-08-19 15:24 GMT+01:00 Jakob Erdmann <[email protected]>: > Hello, > revision 16884 fixes a bug which could explain your symptoms. During > insertion there is a check for maintaining safe distance to the vehicle > behind. The code for determining the required safe distance was returning > an exaggerated value and thus preventing insertion. Please let me know > whether your problem persists with that revision. > (However, note that this revision flags a number of collisions which > previously went unnoticed, see http://sumo-sim.org/wiki/ChangeLog#SVN for > other changes that may apply) > regards, > Jakob > > > 2014-08-19 14:09 GMT+02:00 Jakob Erdmann <[email protected]>: > > Hello, >> if the car in question is on the road in the centralized scenario that it >> should be safe to add it in the segmented scenario. If insertion fails this >> may point to a bug somewhere in the insertion code. If you can provide me >> with a small example I would like to investigate this. >> The safe distance of a car is strongly tied to its braking capabilities. >> (higher decel requires a smaller gap to the leader, lower decel requires >> a smaller gap to the follower) >> It is also strongly tied to the 'tau' value. You could modify these >> temporarily for insertion (since you are using TraCI anyway) and then reset >> them after insertion. >> >> regards, >> Jakob >> >> >> 2014-08-19 13:06 GMT+02:00 Quentin Bragard <[email protected] >> >: >> >> Hi guys, >>> >>> I'm currently facing the same kind of issue while working on a >>> distributed version of SUMO. In the centralised scenario, traffic jam >>> occurs on roads which are situated on borders in the distributed scenario. >>> Therefore, though there is enough place for the cars as shown by the >>> centralised scenario, SUMO refuses to add the new cars arriving on the >>> border. >>> I've tweak SUMO source code by myself a while ago and I was wondering if >>> there would be a way to tweak it to remove/reduce the safe distance >>> required for inserting new car. Any advice about where in the source code I >>> should look to modify this behaviour ? >>> >>> Thanks in advance for your help, >>> Regards, >>> Quentin >>> >>> >>> 2014-07-16 9:36 GMT+01:00 <[email protected]>: >>> >>>> Hi Jakob, >>>> >>>> I've tried a few options, and not had much luck. So, I commented out >>>> lines 699-702, and on the vehicle collision, the sim freezes. Also >>>> tried a few variations of commenting out some additional content, with >>>> same result. >>>> >>>> But, thats not so bad for me, because I use the teleport from traci as >>>> the trigger to tell an agent its crashed. Without the teleport I'd >>>> have to think of some alternative. >>>> >>>> The other problem, is I still don't have much luck getting a quick >>>> re-insert, I seem to need a gap of about 10 seconds. I guess, its not >>>> totally unreasonable, a motorway vehicle travelling at normal speed, >>>> for braking distance to a stationary vehicle. But, would none of the >>>> vehicle parameters affect this? I tried reducing minGap to very small, >>>> didn't make much difference. >>>> >>>> If theres no easy option, no problem, I'll just work the scenario >>>> around having larger vehicle gaps. >>>> >>>> Thanks >>>> >>>> Vincent >>>> >>>> Quoting Jakob Erdmann <[email protected]>: >>>> >>>> > Hello, >>>> > since insertions are only performed if they are deemed safe under all >>>> > circumstances, an insertion speed of zero means that the following >>>> traffic >>>> > needs to have sufficient space to come to a complete stop as well >>>> (when >>>> > starting to brake in the next simulation step). If you find that an >>>> even >>>> > larger gap is required, it might point to a bug in the insertion code >>>> and >>>> > should be investigated. >>>> > However, in your case it sounds as if you would be better of by simply >>>> > deactivating the 'telepor-after-collision' behavior. >>>> > This is something we would like to make configurable eventually >>>> anyway >>>> > (see http://sumo-sim.org/trac.wsgi/ticket/1102) >>>> > Until this is implemented you may comment out some lines in >>>> MSLane.cpp. I >>>> > think 699-702 should be enough. >>>> > regards, >>>> > Jakob >>>> > >>>> > >>>> > 2014-07-13 23:05 GMT+02:00 <[email protected]>: >>>> > >>>> >> >>>> >> Hi, >>>> >> >>>> >> To simulate a crashed vehicle holding up a lane, when I catch a >>>> >> teleported vehicle, I re-insert it at its previous position, with a >>>> >> small route and zero speed. >>>> >> >>>> >> I've set up a dedicated vehicle type for this, mainly so that I can >>>> >> show it in a unique colour, but it does mean I can play with vehicle >>>> >> parameters too. >>>> >> >>>> >> The problem I have, is that I think the 'safe distance' metric might >>>> >> be kicking in but in a more extreme way than usual? For some reason I >>>> >> seem to need about 9 to 10 seconds gap between vehicles, compared to >>>> >> inserting vehicles into the road network normally which is around 2-3 >>>> >> seconds gap. >>>> >> >>>> >> I've tried changing minGap in case this made a difference, but >>>> doesn't >>>> >> seem to. The only unique thing I can think of is that I insert the >>>> >> 'crashed' vehicle back with a speed of 0. >>>> >> >>>> >> The weirder thing, is that if the gap is approx 8 seconds, it seems >>>> >> insufficient, and so a number of vehicles go past the location I wish >>>> >> to re-insert the crashed vehicle, but then all of a sudden it will >>>> >> insert, maybe 2-3 seconds ahead of another vehicle. >>>> >> >>>> >> I'm a bit stuck, the only workaround I have at the moment is to make >>>> >> sure there is a large gap behind any vehicle that is likely to crash, >>>> >> so that it can be inserted, which isn't so realistic. >>>> >> >>>> >> Thanks >>>> >> >>>> >> Vincent >>>> >> >>>> >> >>>> >> >>>> >> >>>> >> >>>> ------------------------------------------------------------------------------ >>>> >> _______________________________________________ >>>> >> sumo-user mailing list >>>> >> [email protected] >>>> >> https://lists.sourceforge.net/lists/listinfo/sumo-user >>>> >> >>>> >>>> >>>> >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> Want fast and easy access to all the code in your enterprise? Index and >>>> search up to 200,000 lines of code with a free copy of Black Duck >>>> Code Sight - the same software that powers the world's largest code >>>> search on Ohloh, the Black Duck Open Hub! Try it now. >>>> http://p.sf.net/sfu/bds >>>> >>>> _______________________________________________ >>>> sumo-user mailing list >>>> [email protected] >>>> https://lists.sourceforge.net/lists/listinfo/sumo-user >>>> >>> >>> >>> >>> -- >>> Quentin Bragard >>> Diplômé de l'INSA de Lyon, Département Informatique. >>> >>> Ph.D Student in PEL, >>> University College of Dublin >>> >> >> > -- Quentin Bragard Diplômé de l'INSA de Lyon, Département Informatique. Ph.D Student in PEL, University College of Dublin ------------------------------------------------------------------------------ _______________________________________________ sumo-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sumo-user
