Implementation considerations

1. High-Level Architecture
   Run both localizers in the system:
   amcl → coarse/global localization
   slam_localizer (e.g. slam_toolbox) → fine/docking localization

2. TF / Pose Switching
   Ensure only one source effectively drives map → odom at a time.

3. Switching Trigger
   Switching would be based region (geofence): robot enters a predefined docking zone (polygon/radius in map frame).

4. Pose Initialization
   When switching AMCL → SLAM:
   Use the current AMCL pose as the initial pose for SLAM localization.
   Wait until SLAM is stable before exposing its TF/pose to the rest of the system.

5. Nav2 Integration
   Nav2 always sees a single, consistent global frame (e.g. map).
   After a switch:
   Trigger a replan or clear costmaps if needed to handle small pose jumps.
   Keep controller paused until a valid pose from the new localizer is available.

Have anyone done this before or working on it? I want help to implement this on my turtlebot3 simulation. Give me suggestions and thank you in advance.