autoware_localization_evaluation_scripts

This package contains some scripts for evaluating the localization of Autoware.

The scripts are used for the result rosbags of localization, particularly the result_bag from driving_log_replayer_v2.

https://tier4.github.io/driving_log_replayer_v2/quick_start/setup/

As a test, execute driving_log_replayer_v2 with the following command:

ros2 launch driving_log_replayer_v2 driving_log_replayer_v2.launch.py \
    scenario_path:=$HOME/driving_log_replayer_v2/localization.yaml

Then, use the scripts for processing the result_bag:

$HOME/driving_log_replayer_v2/out/latest/result_bag

plot_diagnostics.py

ros2 run autoware_localization_evaluation_scripts plot_diagnostics.py \
   <rosbag_path> \
   --save_dir=/your/path (default:rosbag_path/../)

[Example]

$ ros2 run autoware_localization_evaluation_scripts plot_diagnostics.py \
    $HOME/driving_log_replayer_v2/out/latest/result_bag
[INFO] [1740561002.740748735] [rosbag2_storage]: Opened database '$HOME/driving_log_replayer_v2/out/latest/result_bag/result_bag_0.db3' for READ_ONLY.
save_dir=PosixPath('$HOME/driving_log_replayer_v2/out/latest/diagnostics_result')

This command outputs each diagnostic (tsv) and plot result (png).

$ tree $HOME/driving_log_replayer_v2/out/latest/diagnostics_result
$HOME/driving_log_replayer_v2/out/latest/diagnostics_result
├── gyro_bias_validator__gyro_bias_validator.png
├── gyro_bias_validator__gyro_bias_validator.tsv
├── localization__ekf_localizer.png
├── localization__ekf_localizer.tsv
├── localization__pose_instability_detector.png
├── localization__pose_instability_detector.tsv
├── localization_error_monitor__ellipse_error_status.png
├── localization_error_monitor__ellipse_error_status.tsv
├── ndt_scan_matcher__scan_matching_status.png
└── ndt_scan_matcher__scan_matching_status.tsv

0 directories, 10 files

[Example : ndt_scan_matcher__scan_matching_status.png]

extract_pose_from_rosbag.py

ros2 run autoware_localization_evaluation_scripts extract_pose_from_rosbag.py \
   <rosbag_path> \
   --save_dir=/your/path (default:rosbag_path/../)

[Example]

$ ros2 run autoware_localization_evaluation_scripts extract_pose_from_rosbag.py \
    $HOME/driving_log_replayer_v2/out/latest/result_bag \
    --target_topics "/localization/kinematic_state" \
                    "/localization/pose_estimator/pose_with_covariance"

This command outputs tsv files for each pose.

The file names are the topic names with “/” replaced with “__”.

$ tree $HOME/driving_log_replayer_v2/out/latest/pose_tsv
$HOME/driving_log_replayer_v2/out/latest/pose_tsv
├── localization__kinematic_state.tsv
└── localization__pose_estimator__pose_with_covariance.tsv

0 directories, 2 files

compare_trajectories.py

ros2 run autoware_localization_evaluation_scripts compare_trajectories.py \
   <tsv_file_subject> <tsv_file_reference>

[Example]

File truncated at 100 lines see the full file

Changelog for package autoware_localization_evaluation_scripts

0.4.0 (2025-11-16)

• feat(autoware_localization_evaluation_scripts): add new diagnostics to check and refactor diagnostics_flag_check (#310)

  • fix process when there is no DiagnosticsFlagCheck in scenario
  • fix typo
  • add diagnostics flag check_is_initial_pose_reliable
  • refactor script
  • add debug print
  • add nvtl and localization_error_monitor diags
  • make time expression richer
  • make sec and nanosec explicilty int
  • fix large int casting
  • add pose_instability
  • Update document
  • make the code clean
  • fix spelling error

  * remove unnecessary print ---------

• chore: update maintainer (#304)

  * chore: update maintainer remove Ryu Yamamoto

  * chore: update maintainer remove Kento Yabuuchi

  * chore: update maintainer remove Koji Minoda

  * chore: update maintainer remove Shintaro Sakoda ---------

• fix(autoware_localization_evaluation_scripts): fix diagnostics flag check expection handling (#303)

  • fix process when there is no DiagnosticsFlagCheck in scenario

  * fix typo ---------

• feat(autoware_localization_evaluation_scripts): detect rise/fall flags of localization diagnostics (#298)

  • Enable to read diagnostics flags
  • make the constructor of OverallCriteriaMask robust to unknown conditions
  • make input of diagnostics_flag_check simple and make large int value inputs robust
  • add document for diagnostics_flag_check
  • fix cspell, add explanation

  * fix docs mistake ---------

• feat(autoware_localization_evaluation_scripts): create masks for each criterion (#297)

  • add mask to evaluate each criterion
  • minor fix
  • remove print debugs

  * fix spelling mistakes ---------

• Contributors: Motz, Taiki Yamada

0.3.0 (2025-08-11)

• fix: change the criteria for localization (#267)

  • Added the ndt filter
  • Fixed the threshold from 1% to 5%

  * Removed the redundant line ---------

• feat: add acceleration evaluation (#265)

  • Added acceleration evaluation
  • style(pre-commit): autofix
  • Added a blank
  • Fixed comments
  • Refined calc_acceleration_diff.py

  * style(pre-commit): autofix ---------Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

• chore: change the maintainer of [autoware_localization_evaluation_scripts]{.title-ref} (#266)

  • Changed the maintainer of [autoware_localization_evaluation_scripts]{.title-ref}

  * Fix

File truncated at 100 lines see the full file