-
Notifications
You must be signed in to change notification settings - Fork 129
Consideration of introducing an s/t-coordinate system #570
Copy link
Copy link
Closed
Labels
AnalysisAn issue or MR that needs expert analysis to determine what to do nextAn issue or MR that needs expert analysis to determine what to do nextConceptAn issue that is being detailed out through expert discussion and offline workAn issue that is being detailed out through expert discussion and offline workRoadModel4.0A label to collect road model topics for OSI 4.0A label to collect road model topics for OSI 4.0
Milestone
Metadata
Metadata
Assignees
Labels
AnalysisAn issue or MR that needs expert analysis to determine what to do nextAn issue or MR that needs expert analysis to determine what to do nextConceptAn issue that is being detailed out through expert discussion and offline workAn issue that is being detailed out through expert discussion and offline workRoadModel4.0A label to collect road model topics for OSI 4.0A label to collect road model topics for OSI 4.0
Describe the feature
From the side of some traffic participant modellers the request was raised to introduce s/t-coordinate systems because it would make certain modelling tasks much easier. Before making an implementation suggestion I would like to assess potential advantages and disadvantages of the concept, as well as discuss if it would make more sense on a lane or a road basis.
Potential advantages are seen in helping with the following tasks:
Of course all those calcuations could be done with x,y,z coordinates but it is more difficult and many simulators have s/t coordinates anyway.
Potential disadvantages:
Furthermore the s/t coordinate system could help in referencing a road surface description.
Describe the solution you would like / Describe alternatives you have considered
I could imagine two solutions and I don't know yet which is the best:
Extend an s-coordinate to each lane, i.e. to lane boundary points (or centerline points, if this concepts stays in 4.0). Here I can think of 2 issues: First, if the coordinate system is just the linear connection of points, not every point can be represented in s/t due to kinks:

Could be solved if polynoms where fittet but then the question is which polynom and through how many points. The second issue might be, that the same s-value for two adjacent lanes would NOT mean that vehicles were next to each other, because the inner-radius of a curve is smaller.
Introduce a data type "Road" and introduce an s-coordinate to each road. Here the first issue is equivalent while the second isn't there. Of course, some effort would be needed to define a road message.
Describe the backwards compatibility
Additional context