Draft for discussion: Surface lines to model road shapes and heights

[arnedueselder]

Surface lines concept as proposed in the OSI Road Network Model Group

First draft on how to implement the surface line concept in OSI. As discussed, the approach follows the concept of Vector3D line segments that form a lane and are used to retrieve height information at different parts of that lane.
Speed bumps etc. could be modelled with more points, while almost flat surfaces would require none or almost none surface lines at all.

Height information of lane boundaries should be also included. The concept should only be used for a very rough description of the height profile and not used for a very detailed description of the road surface.
It would replace the currently implemented centerline for now (but not necessarily). The intend is to separate physical properties from logical information. It would still be possible though to place a surface line in the center of the lane to keep that information.

Concept will be expanded to be used in junctions, but is still open for discussion within the concept group.

[arnedueselder]

Discussions from Meeting at 22.11.2021:

• If more than one surface line exists they should maybe contain an equal number of points along the lane to make indexing positions easier.
• That would probably mean also that either lane boundaries should be sampled the same or that a lane boundary also contains its own surface line with the appropriate number of total points.
• Approximation definition for the current centerline should be expanded and applied also for height information to minimize errors between to points.
• Surface lines possibly are better placed within logical lanes instead of lanes Logical Lanes in OSI #599

Enums for new variables should be added at the end to be backwards compatible.

[thomassedlmayer]

Maybe we would also like to describe the transition from a road surface to an elevated sidewalk with this approach.

In order to describe the curb elevation using the current surface concept we would have to define another right- or left-adjacent surface line (or lane boundary) which is located above the bottom lane boundary. Is this a good way to define it?

Also, there is a height parameter for each BoundaryPoint of a lane boundary, which already defines some vague height of curbs (height ≠ z-coordinate). There is no definition of its "direction vector" though (e.g. perpendicular to some surface that does not exist yet?). We should probably think about some precise specification there. Maybe it could be used for the surface concept?

[clemenshabedank]

Closing in favor of #619.