Hello, I recommend setting a minGapLat value that prevents interaction between vehicles driving next to each other in the center of their lanes. To do this, it is sufficient to set minGapLat <= laneWidth - vehWidthMax (0.7 in your scenario).
Explanation: the current model ties the actual gap requirement to the vehicle speeds and when these fluctuate just a little bit, the vehicles adjust their gaps accordingly. As the model continues to evolve I may be able to find a better way to prevent these oscillations so this is not the final word. regards, Jakob 2017-08-08 13:48 GMT+02:00 Kaths, Jakob <[email protected]>: > Hi Jakob, > > > > thank you for looking into that. The problem seems to be minimized, but, > unfortunately, when raising the demand, I can still observe the oscillating > behavior with the current nightly snapshot (r25506). I added the same > scenario with higher demand: https://syncandshare.lrz.de/ > dl/fiErEWR5TJnGw2hqCzygZRMW/minimal_test_fishtail_higherdemand.zip > > > > When looking at AB_counter_truck.36 at simulation second 76.2 the blinkers > and angels still show frequent changes. Should I just fix the parameters > and if so, how is the relation between the different values to prevent this > behavior? > > > > Thanks again for the great work, the sublane feature is very helpful! > > > > Jakob > > > > > > *Von:* Jakob Erdmann [mailto:[email protected]] > *Gesendet:* Montag, 7. August 2017 14:55 > *An:* Kaths, Jakob <[email protected]> > *Cc:* [email protected] > *Betreff:* Re: [sumo-user] Swerving / fishtailing of vehicles when using > sublane model > > > > Thank you for providing the example scenario. The issue ( > http://sumo.dlr.de/trac.wsgi/ticket/3333) is fixed in revision [25500]. > > The relevant aspect of your scenario for exposing this problem was a > mismatch between lane width, vehicle width and the minGapLat attribute > (default 1.0m). In your scenario, the vehicles were unable to maintain > their minimum preferred lateral gaps even when driving in the center of > their lane. Setting vType attribute minGapLat="0.8" would also mitigate the > issue. > > regards, > > Jakob > > > > 2017-07-25 10:48 GMT+02:00 Kaths, Jakob <[email protected]>: > > Hi, > > > > I hope I prepared a meaningful and minimal enough example. It can be > downloaded here: https://syncandshare.lrz.de/dl/fiMpaoK6ajFTxaMLLcs5czyX/ > minimal_test_fishtail.zip > > > > For example AB_counter_car.3 leaves the network straight on the second > lane, whereas AB_counter_car.27 alternates between 203.88 degrees and > 205.86 degrees from step 69.1 on. > > The alternating indicators can be seen on some of the trucks. With curved > networks and more variable traffic, the effects become more obvious though. > > > > I hope that helps, let me know, if I can provide anything else. > > > > Thanks again > > > > Jakob > > > > *Von:* Jakob Erdmann [mailto:[email protected]] > *Gesendet:* Dienstag, 25. Juli 2017 14:36 > > > *An:* Kaths, Jakob <[email protected]> > *Cc:* [email protected] > *Betreff:* Re: [sumo-user] Swerving / fishtailing of vehicles when using > sublane model > > > > Hello, > > fluctuation decisions indicate a bug in the sublane lane-change model. You > can help to get this fixed by providing a minimal example scenario (the > fewer vehicles and edges the better). > > regards, > > Jakob > > > > 2017-07-25 6:10 GMT+02:00 Kaths, Jakob <[email protected]>: > > Hi, > > > > sorry to warm this topic up again. I’ve been looking a bit more into it > again and I definitely see some improvement, but the behavior of the > sublane model with 0.1 s step still leads to a frequent change in the sign > of the yaw rate of vehicles. Furthermore, for some vehicles, there seem to > be fluctuating decisions about lane changes, so each step they redecide and > the indicators blink right and left alternating each step. > > Is this still considered a bug or just the nature of the sublane > lane-change model? If the latter is the case, are there suggested parameter > sets to damp this behavior? > > > > Thanks a lot for your suggestions > > > > Jakob > > > > *Von:* Kaths, Jakob > *Gesendet:* Freitag, 14. Juli 2017 13:18 > *An:* [email protected] > *Cc:* 'Jakob Erdmann' <[email protected]> > *Betreff:* AW: [sumo-user] Swerving / fishtailing of vehicles when using > sublane model > > > > Sorry, I should have tried the latest developer version first. Looks a lot > better on first glance! Thanks! > > > > *Von:* Jakob Erdmann [mailto:[email protected] > <[email protected]>] > *Gesendet:* Donnerstag, 13. Juli 2017 14:10 > *An:* Kaths, Jakob <[email protected]> > *Cc:* [email protected] > *Betreff:* Re: [sumo-user] Swerving / fishtailing of vehicles when using > sublane model > > > > Hello, > > this is due to a bug which has been fixed in the latest development > version. > > regards, > > Jakob > > > > 2017-07-13 8:05 GMT+02:00 Kaths, Jakob via sumo-user < > [email protected]>: > > Hi everybody, > > I know there have been some bugfixes in this area lately, but still I > observe somewhat odd behavior when using the sublane model. Especially > longer vehicles (trucks, busses) in some parts of my network seem to > fishtail/swerve for no apparent reason. They perform oscillating angle > changes, which makes them "wiggle" without actually changing lanes. I set > the lateral resolution to 0.5, the step size to 0.1 and I'm using the SUMO > 0.30.0 release. > Are there any hints on how to avoid this behavior? I tried adjusting the > lateral speed (to 0.1), but it doesn't help. > > Thanks! > > Jakob > > -- > Dipl.-Ing. Jakob Kaths > > Technische Universität München > Ingenieurfakultät Bau Geo Umwelt > Lehrstuhl für Verkehrstechnik > Univ.-Prof. Dr.-Ing. Fritz Busch > > [email protected]<mailto:[email protected]> > www.vt.bgu.tum.de<http://www.vt.bgu.tum.de/> > > > ------------------------------------------------------------ > ------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > _______________________________________________ > sumo-user mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/sumo-user > > > > > > > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ sumo-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sumo-user
