Hi!
Did anybody figure out how “exactly” the benchmark calculates the velocity value ?
Both the paper from the commonroad team and the original source are essentially:
sum(x = 0 to n) ((goal_velocity - current_velocity(x)) ** 2)
(sum over all velocity offsets in regards to the goal squared)
But the velocity in the goal is typically a interval. (start to end)
I can’t seem to approximate the value of the benchmark with my own calculation correctly.
Neither start, end nor average of the interval seem to be right.
The same goes probably for orientation, which is also represented with it’s own value in the benchmark
If anybody got the value right or someone from the team could explain the implementation of the benchmark I would be thankful.
-commonvehicle