LiDAR localization becomes unstable

1. Symptoms

• The pose jumps or oscillates indoors.
• Particles spread and split.
• Localisation holds in some rooms and fails in others.

2. Possible causes

• The environment differs from the map (moved furniture, new objects).
• Symmetric or featureless places (corridors, halls).
• Motion distortion in the scans while turning.
• Wrong sensor frame, range limits or scan rate.
• Odometry noise parameters too small or too large.
• Poor initial pose.

3. Measurements to collect

• Particle spread or covariance over time.
• Match score against the map if available.
• Scan overlaid on the map at the time of failure.

4. Diagnostic commands

Commands are not executed by the site’s automated tests (they need a ROS 2 Jazzy installation or the robot).

ros2 topic echo /particle_cloud --once
ros2 topic echo /amcl_pose --field pose.covariance
ros2 topic hz /scan

5. Step-by-step investigation

1. Overlay the scan on the map at the failure.
2. Check the places where it fails for symmetry or change.
3. Check scan frame, range and deskew.
4. Check odometry noise parameters and the update thresholds.
5. Check the initial pose handling.

6. Corrective actions

• Update the map or add landmarks.
• Add constraints (markers, other sensors) in symmetric places.
• Deskew scans and fix frames.
• Tune odometry and update parameters.

7. Validation after the fix

• Localisation stays within the reported covariance over repeated runs through the problem places.

References

• Nav2 AMCL README and default parameters (jazzy): nav2_amcl, nav2_params.yaml, read 2026-11-06.