Centerline for other lane types?

[clemenshabedank]

After a discussion in Harmonization WP we decided to open an issue on the concept of the centerline in OSI to clarify the usage and discuss if the description of a centerline needs to be changed. The current definition can be found in https://github.com/OpenSimulationInterface/open-simulation-interface/blob/master/osi_lane.proto, line 158-205. There, the centerline is explained in detail. Among other things, it states

• "The centerline describes the middle of the lane."
• "The center line is the line that a typical vehicle follows more or less (depending on the situation, a little more to the left or right of the center line"

It also explicitly states

• "Intersections and non-driving lanes do not have a center line."

The question that arose is, if the centerline might be useful for soonish to be introduced subtypes of TYPE_NONDRIVING, especially SUBTYPE_SIDEWALK and SUBTYPE_BIKING. Not representing real physical assets, I suppose the centerline is already used for the purpose of guiding the ego or other traffic participants. So it could also help to guide bicycles or pedestrians. Also as already stated in the proposal (see requirement 1.3.10 point 2 from https://code.asam.net/simulation/proposal/osi), there could "be a centerline (typical driving route) on junctions, favorably for different vehicle sizes", so for the TYPE_INTERSECTION.

Any opinions to that question? Should the concept of the centerline be allowed for other lane types than TYPE_DRIVING? Or should we even think of a totally different concept? I think the explanation shall be rewritten a little anyways, because I think it currently is not very specific.

[engelben]

We had a general Domain Modeling discussion for ASAM yesterday - we decided to create a subgroup that involves experts from all the OpenX Projects for two topics (to start):

1. Roads & Lanes
2. Traffic Participants.

The focus of these subgroups is to discuss exactly things like this, with a particular emphasis on cross project alignment and understanding the needs of the individual standards.

This is being coordinated by the OpenXOntology Project so I will ask the responsibles to include you @clemenshabedank

[lemmer-fzi]

From a traffic participant point of view it would be helpfull to have a "virtual centerline" definition on junctions for each linking lane.

[clemenshabedank]

Concern from a colleage: it would be very desirable in my opinion, if there were not only centerlines in the crossing area, but the complete lanes from OpenDRIVE (so with Boundaries). Reason: There exist usecases where the intersection lanes are used to define conflict areas, i.e. areas where several lanes overlap and where you have to wait because of right of way etc. I imagine this to be difficult only with Centerline. Sould be regarded in the discussion.

[jdsika]

We at the time decided to just display lines that are visible to the sensor on roads and crossings. The centerline was always subject to discussion and we kept only the information on "what road is connected with each other" to evaluate possible driving directions.

[clemenshabedank]

Decision for now: The harmonization workinggroup decided not to touch the centerline definition for non-drivable lanes. Reasons: • Would result in significant increase of data • Centerline can be deduced based on the laneboundaries • The centerline concept should be revised for the next major release

[clemenshabedank]

Status on this topic (05/2022) in the Road Model Group: Centerline is already problematic to use for driving paths (although used a lot). Therfore, it is seen problematic to introduce it for non-driving lanes. Potential deprecation of centerline was discussed with no conclusion yet.