localization-recovery

To recover AMCL localization when navigation breaks down, trigger global relocalization via the /reinitialize_global_localization service, monitor pose covariance, or manually reset the pose using RViz 2D Pose Estimate.

AMCL divergence is detected by monitoring high covariance in /amcl_pose, which indicates pose uncertainty. This degradation causes map-odom inconsistencies, requiring global relocalization or particle re-seeding to re-converge on the map.

Automate AMCL recovery by integrating behavior-tree workflows that monitor /amcl_pose covariance and trigger service-based relocalization. This provides safe, repeatable recovery when localization failures occur.

Yes, pose persistence across reboots avoids cold-start localization when the robot has not moved. This prevents initial localization failures by retaining the last known accurate pose instead of relying on default startup.

Manually reset AMCL initial pose using RViz' 2D Pose Estimate tool or by publishing directly to the /initialpose topic. This provides an immediate pose reset when automated global relocalization is insufficient.

Tune AMCL particle count when covariance monitoring indicates persistent high uncertainty during recovery. Optional particle-count tuning helps the filter re-converge and maintain robust localization during global relocalization procedures.