support bidi edge interaction for all vehicles

[namdre]

context: bicycle simulation where a wide road is used in both directions across the whole width.

netconvert:

• define forward and backward edge with the same (reversed) geometry and set edge attribute bidi="true" (same idea as in railway simulation)
• preserve bidi attribute when loading .net.xml with netconvert
• add option --bidi.guess moved to allow detection of bidirectional lanes based on overlap #13546

netedit

• make 'bidi' into computed/overrideable edge attribute (type string) in netedit
• visualization when 'spread' is set

sumo

• identify overlapping lanes and assign the corresponding value to be returned by MSLane::getBidiLane (widths must match)
• map vehicles from the bidi-lane as partial vehicles into the current lane so they are recognized during MSVehicle::planMove and MSLaneChanger::change
• refactor code that detects opposite vehicle direction (must check lcModel->isOpposite as well isOnBidi)

sumo-gui

• visualization when 'spread' is set

Benefits:

• permits simulation of narrow roads, ""two-way left turn lane" and "suicide lane" arrangements even without sublane model.
• high fidelity bicycle simulation with sublane model

[yunpangfloetteroed]

The directory names of the tests represent the issues which should be checked.

[yunpangfloetteroed]

Vehicle still wiggles especially when the second bike is passing by and the vehicle makes an emergency braking which is not expected (he should be able to see the second bike coming eariler).

The sumo version v1_13_0+0846-9b68eadfd9
bikePassingWithVehWiggle.zip
6

[yunpangfloetteroed]

With latAlignment="right" only the eastbound bike changes his side for overtaking. The westbound bike still keeps on the right side. It causes a collision, but SUMO does not report it in the sumo-gui's information window.
bikeNotAvoidConflictWithBike_delayedTeleporting.zip

[yunpangfloetteroed]

Regarding my previous message, it is also unexpected that the vehicle makes an emergency braking when meeting the bikes. The vehicle should be already aware of the bikes on the same link where the vehicle is on.

It also seems that bikes only try to overtake the repsective vehicle from their left side. They could also do overtaking from the right side in the reality. I would say (my own experiences), it depends more or less on the available space around the object, which is overtaken, and the size and speed of the repsective object. Would it already be considered in SUMO?

[namdre]

There was a change in this regard in 7221494. Do you still see a problem with any of the existing tests? If not can you provide a new test case?

[yunpangfloetteroed]

Regarding overtaking on the left hand side it can still be seen in the existing test ~\sumo_git\tests\sumo\sublane_model\bidiEdge\bikeFaultCollision

[yunpangfloetteroed]

Yes, now both of them pass by the vehicle on the right hand side.

When the cyclist, in the aforementioned test, is going to approach/pass the vehicle from the other direction, he/she is still speeding up, like an overtaking action. But, I think, he/she should take an action more like "conflict avoiding" than "overtaking". So, the repsective speed may be actually getting slower or keeping the same. What do you think?
Are both cases dealed as overtaking action in SUMO currently?

[namdre]

Actually, the speeding happens because the cyclist is below the desired speed. It is not directly related to overtaking.

[yunpangfloetteroed]

you mean the pre-defined max speed. right? it is actually not desired speed, but technical speed ability, as you wrote in #11102. Perhaps this could be changed after your implementation in #11102.

[namdre]

It would be more accurate to say that the maxSpeed is currently used to model the desired speed since we don't have a distinct value for desiredMaxSpeed yet and bicycles have no relevant technical limit in this scenario.

[yunpangfloetteroed]

Yes, you are right with this point. Sorry, I did not make my point clear. The issue I would like to point out is that the cyclist should not try to speed up for reaching his/her desired speed, under such condition since he/she is actually under some "pressure" for safely riding through the narrow space between him/her and the vehicle (my own experience). In my case, I might (1) even stop for let the vehicle pass firstly or (2) continue and speed up when the vehicle stops and wait for that I pass by. The current situation is more like a situation with autonomous bike. And, the bike is 100% sure to run through the narrow space, while the vehicle aslo runs at the same time.

[felixfroehling]

Hey, is there any way I can already test the current development status?
For example in a nightly snapshot version?
And do you guys know, when the feature will be officially released?

[namdre]

Hey, is there any way I can already test the current development status?

You can test the current features with 1.14.1 with examples from https://sumo.dlr.de/extractTest.php?path=sumo/sublane_model/bidiEdge

[yunpangfloetteroed]

Unable to execute isBidi in netedit also when the box is not shown grayed out. Together with the attached example file the situation is described as below

• the links (E1, -E1, 977388750, -977388750, 371502353#0, -371502353#0) could be set via isBidi
• These links were reset by klicking isBidi again
• Then it does not work when trying to set these links as bidi-links.
  problem.net.xml.zip