| Re: [sumo-user] Multi-lane homogeneous congestion |
|
Dear Jakob, I can confirm that more lane changes occur from the middle to the left after decreasing
lcSpeedGainUrgency. Thanks also for the advice about the acceleration lane length. More lane-changes can take place by making it longer. I see that there are some parameters for cooperative speed adjustments, such as
lcCooperativeSpeed, but mostly for strategic lane change requests. In the description of lcSpeedGainUrgency, it stated “triggers speed adaptation in neighboring (obstructing) vehicles”. Does this mean that there is actually some speed adaptation mechanism on the target lane when there is speed gain lane change request? Thanks in advance for the answer! Best, Ying-Chuan From: Jakob Erdmann <namdre.sumo@xxxxxxxxx> Hello, this is as known issue and the current approach for solving it, is via https://github.com/eclipse-sumo/sumo/issues/1145 You'll get a somewhat more even distribution of speeds by setting lcSpeedGainUrgency="3" It also helps to remove actionStepLength="1.0". By getting more lane-changing attempts you see more cars moving from the middle to the left lane. Finally, your hard-coded acceleration lane length of 100m is too short. Typical lengths on German and Swiss motorways is 250m. Using all of these the original speed distribution: <edge id="E6"> changes to <edge id="E6"> The implementation of #1145 is not complete. The biggest obstacle is https://github.com/eclipse-sumo/sumo/issues/16302 In dense traffic, the speed estimation sometimes fails and vehicles on the slow lane do not collect enough incentive for attempting to change which is not how the model is supposed to work. regards, Jakob Am Do., 22. Okt. 2026 um 17:11 Uhr schrieb Ni Ying-Chuan via sumo-user <sumo-user@xxxxxxxxxxx>:
|