path-tracking-metrics

Record /odom and /plan topics with ros2 bag record, then compute the lateral distance from each odometry position to the nearest point on the planned path. The included Python script automates this using the rosbags library and reports mean, max, and standard deviation.

For indoor navigation, cross-track error below 0.10 m is excellent and suitable for narrow corridors, while 0.10-0.20 m is normal. Values above 0.30 m indicate significant wandering, usually caused by misconfigured controller critic weights.

Run ros2 bag record with the /odom, /plan, /local_plan, /cmd_vel, /tf, and /tf_static topics during a typical patrol run. Including the tf topics allows later replay and transform lookups during offline analysis.

High heading error means the robot is skewed relative to the path tangent, typically because PathAngleCritic is too weak. Increase PathAngleCritic.cost_weight or lower max_angle_to_furthest in the Nav2 controller configuration.

A path length ratio above 1.3 means the robot traveled much farther than the planned path, indicating major detours, oscillations, or repeated recovery behaviors. Check the Oscillation critic for DWB or TwirlingCritic for MPPI controllers.

Yes, the analysis script uses the rosbags Python library, which reads rosbag2 files standalone without a ROS 2 installation. Install it with pip alongside numpy and run the script directly against the bag directory.