Result Comparison

The metric apps are designed to compare one trajectory to a reference. While this simplifies the usage, the comparison of different results is usually also interesting.

With evo, the result comparison is a separate step. This has the advantage that you can compare the results of a SLAM algorithm X to another algorithm Y without having to run the metrics all over again. And the comparison is independent of the original trajectory file format.

--save_results

The workflow is that you run the metrics for each trajectory and save the results individually in files by specifying the --save_results argument in the metric app. Have a look at the README.md for an example.

The "result files" are serialized Result objects stored in a zip format that can be read by evo_res.

evo_res

You can load multiple of those result files with evo_res to compare the different results. To avoid wrong comparisons, it will warn you if it thinks that you compare files that were generated by different metrics (unless --no_warnings is specified).