Planning evaluation results on Nuscenes

[aurora-xin]

Thank you for this nice work!
I would like to know how to obtain the planning evaluation results shown in Table 1 of the paper.
Are they computed by averaging all ade_results.jsonl files for each scene equally? Any guidance would be appreciated!

[ShuoXing98]

Hi @aurora-xin ,

For the results presented in Table 1, we first conduct the experiments on the scenes of the eval split of the nuscenes Full dataset (v1.0, us). A prediction is classified as a failure if the L2 norm exceeds 10 within the first second of the predicted future trajectory. For predictions that do not fail, we compute and report the average L2 norm error.

[kooWZ]

@ShuoXing98 Could you release the code for evaluation? (i.e. calculating the final average L2 norm)