Skip to content

Latest commit

 

History

History

eval

KISS-ICP Evaluation

In this folder you can find a few Python notebooks to run a complete evaluation of KISS-ICP without much effort.

Datasets evaluated

NOTE: In this repository we add the empty notebooks to use it for running experiments on your local machine. If you want to see the results of these experiments you can click on any click below or directly access the already runned notebooks here

Where is the NCLT dataset?

We discourage using NCLT to evaluate odometry systems: misalignments in the ground truth poses, missing frames, and inconsistencies in the data make the evaluation of odometry systems on such a dataset not a good evaluation tool from our perspective. However, we provide the results here for the sake of completeness in the manuscript.

What about all the other datasets?

Here we just want to add an easy evaluation bench for the KISS-ICP pipeline. Much more datasets are currently supported by our system and more are to come, an in-complete list of such datasets includes:

  • apollo
  • boreas
  • kitti
  • kitti_raw
  • mulran
  • ncd
  • nclt
  • nuscenes
  • paris_luco
  • rosbag
  • rosbag2
  • generic folders containing files with extensions:
    • bin
    • pcd
    • ply
    • xyz
    • obj
    • ctm
    • off
    • st

What about all the other experiments from the paper?

The experiment missing from this evaluation set are those whe require to modify the C++ core library, therefore we could not add Python notebooks for those experiments, but should be easy to run them once the full source code is available.