In quantifying the performance of my (MPC-)motion planner, I intended to utilize the CommonRoad IO framework.
In the paper introducing CommonRoad’s benchmarking capabilities, on page 5, one can find the following mention of papers that published the weights used in their cost-functions:
The cost functions mentioned here – as originating from the respective cited papers – have been used to form some of the standard cost-function configurations that can be chosen from when using the CommonRoad framework.
What I would image to be of great benefit, but am afraid I could not find anywhere, would be the actual optimal cost that was realized by the respective methods mentioned in the screenshot on certain CommonRoad scenarios.
Am I correct in observing that although their cost-functions form the basis of the CommonRoad framework, the methods themselves have not been applied to the CommonRoad scenarios yet, thereby not generating literature-based benchmarks?
Or is there actually a place where the methods in these papers (or related optimal control literature) have been applied to CommonRoad scenarios with the generation of benchmark values as a result?
I am looking for the cost realized by some optimal-control based methods from literature, in summary: are you aware of such benchmarks having already been created on the CommonRoad framework?
I am aware that many student participants have benchmarked their solutions, which I found in the leaderboard, but I seem to understand their methods are not publicly documented and are also often based on AI?