Thank you for the example. I now understand the situation you are referring to. The behavior occurs because the lateral movement during lane changing is independent of the apparent "lateral" movement that occurs on the zipper merge. Once the vehicle comes out of the zipper merge it finds itself in the middle of the started lanechange maneuver and simply completes it. The problem is more severe if there is no further lane to change to: https://github.com/eclipse/sumo/issues/12663 Thank you for bringing this to my attention.
Am Do., 16. Feb. 2023 um 21:22 Uhr schrieb Jordan Poots < [email protected]>: > Hi Jakob, > > I have attachd the appropriate files for lane hopping and removed > therandom seed so the default seed is being used. I will work on > reproducing the failed collision detection but this is more time consuming. > > Kind regards, > > Jordan > ------------------------------ > *From:* sumo-user <[email protected]> on behalf of Jakob > Erdmann <[email protected]> > *Sent:* 16 February 2023 13:24 > *To:* Sumo project User discussions <[email protected]> > *Subject:* Re: [sumo-user] Lane hopping and emergency braking at zipper > > Thanks for the examples, unfortunately, I cannot run them because they > make use of sumo option --random. (The traci commands no longer match the > required simulation state, i.e. vehicle ids are not present at the assumed > time). > If you want to have randomness in a reproducible way you need to use > option --seed SEED with different values of SEED. Then you can send me the > seed that shows the relevant behavior. > > regards, > Jakob > > Am Mo., 13. Feb. 2023 um 18:21 Uhr schrieb Jordan Poots < > [email protected]>: > > Hi again! > > I have managed to reproduce the collision detection failing and haev > attached the files to recreate it. The image below shows the GUI message > for the collision traci didn't detect. > > > Kind regards, > > Jordan > ------------------------------ > *From:* sumo-user <[email protected]> on behalf of Jordan > Poots <[email protected]> > *Sent:* 13 February 2023 16:28 > *To:* Sumo project User discussions <[email protected]> > *Subject:* Re: [sumo-user] Lane hopping and emergency braking at zipper > > Hi Jakob, > > I have produced a simlified example which reproduces the lane hopping. In > this example there is a continuous lane change with no sublane model but as > far as I can tell this problem presents in the sublane model, basic > continuous lane changes and instantaneous changes. This problem arises with > both the laneChange and laneChangeRelative commands. > > The log file for collisions not being detected is more difficult as this > is a lot more random but I will try and generate an example for it. > > Kind regards, > > Jordan > ------------------------------ > *From:* sumo-user <[email protected]> on behalf of Jakob > Erdmann <[email protected]> > *Sent:* 06 February 2023 13:25 > *To:* Sumo project User discussions <[email protected]> > *Subject:* Re: [sumo-user] Lane hopping and emergency braking at zipper > > Without a traceFile I cannot replicate the problem. Simply provide a trace > that is long enough to cover the problematic situation (it will contain all > traci calls that were sent from your code and thus also replicate your > randomness on my end). > > Am So., 5. Feb. 2023 um 02:33 Uhr schrieb Jordan Poots < > [email protected]>: > > Hi Jakob, > > Upon review the vehicles are not stopping but are failing to finish their > continuous change due to the lane ending. This while not ideal, isn't as > bad as what I said earlier. > > My major problem now is lane skipping around the merging zone. The > vehilcle is only allowed to switch lanes on the "merge" edge which has 3 > lanes. However, it seems to be hopping over two lanes when a lane change is > commanded at the zipper (or perhaps right before it?). I have already tried > to limit this by including a position condition. It should not be able to > change lanes at the junction. > > On this same project I also have an issue with TraCI not detecting > collisions which the simulator is. > > I fear a trace file may not show case the problem as I am using a > reinforcement learning model to respond dynamically to the environment > which contains randomness. I have however, included the scenario. The > command being used is a simple traci.vehicle.changeLane(id, 1, 10). > > Kind regards, > > Jordan > ------------------------------ > *From:* sumo-user <[email protected]> on behalf of Jakob > Erdmann <[email protected]> > *Sent:* 30 January 2023 11:04 > *To:* Sumo project User discussions <[email protected]> > *Subject:* Re: [sumo-user] Lane hopping and emergency braking at zipper > > Technically, there should never be a lane-end warning if the zipper > connections are set up correctly. It's hard to tell what's going on without > observng running traci code. > Please put together a minimal example that triggers the error message a > package the scenario files together with a traci-traceFile: > https://sumo.sourceforge.net/docs/TraCI/Interfacing_TraCI_from_Python.html#generating_a_log_of_all_traci_commands > > Am So., 29. Jan. 2023 um 02:06 Uhr schrieb Jordan Poots < > [email protected]>: > > Hello, > > I am using a zipper lane driven by TraCI controlled vehicles. This zipper > lane reduces three lanes to two. The lane changes are continuous. The TraCI > vehicles can merge at any point on the zipper lane or simply follow the > lane and merge as it ends. Once the vehicles have merged, control of them > is returned to SUMO. > > I find when vehicles choose to change lanes late on the zipper lane, they > have two problems. > > 1. Occasionally, they will emergency brake. I receive an error message > about the lane ending. I assume this is in reference to either the to or > from lane ending as the moving vehicle moves onto the zipper junction > during the lane change. > 2. They occaisonally cross two lanes instead of the intended one and > thus end up on the left lane on the reduced side instead of the right. > > How might I configure my setup to avoid these issues? > > Thanks, > > Jordan > > _______________________________________________ > sumo-user mailing list > [email protected] > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user > > _______________________________________________ > sumo-user mailing list > [email protected] > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user > > _______________________________________________ > sumo-user mailing list > [email protected] > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user > > _______________________________________________ > sumo-user mailing list > [email protected] > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user >
_______________________________________________ sumo-user mailing list [email protected] To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
