state-estimator-evaluate-bags

To evaluate state estimator performance, this Skill replays recorded ROS2 rosbag2 MCAP datasets through the mole_estimator, records processed evaluation bags with state and graph information, and generates per-bag performance metrics.

Yes, you can compare state estimator metrics between different runs using the built-in comparison tool, which analyzes differences between two runs to highlight improvements or regressions for tuning.

Offline bag evaluation helps tune turn-joint filtering and base velocity smoothness by running an analyzer on reprocessed bags to generate a markdown report and summary table of per-bag performance metrics.

The best way to automate mole_estimator evaluation is using a Skill that replays `*_sensors` MCAP bags through the estimator, records eval bags, and updates the metrics markdown and summary table automatically.

No specific dependencies are required to run state estimator bag evaluation beyond the standard environment, as this Skill operates independently without external package requirements.

State estimator performance changes across different sensor bags because the offline analyzer generates per-bag metrics reflecting varying dataset conditions, allowing you to identify specific tuning needs for each scenario.