Python evaluation pipeline to analyze the localization and drive/gait of vehicles/quadrupeds in Autoware.
- statistic analysis of the gait/drive with IMU and Twist data
- localization analysis with NDT and EKF performance parameter
- for Simulation only: localization evaluation with ground truth data from the Simulator (e.g. Isaac Sim)
- creation of correlation matrices, evaluation tables and other plots for visualization
- option for zero-phase-filtering all input data
The following topics need to be in the rosbag:
- /diagnostics (for Mahalanobis distance evaluation, the value needs to be added in the autoware code to the diagnostics topic)
- IMU: /sensing/imu/imu_data
- Twist: /sensing/vehicle_velocity_converter/twist_with_covariance
- TF: /tf (for estimation error, ground truth and estimated pose tf need to be recorded)
