Loop closure handling in RACER PLANNER

[zeeshanns]

I have a query about managing pose jumps in exploration planning when using SLAM. Imagine a robot equipped with an exploration planner and a separate SLAM module. During exploration, if the robot returns to its starting point (Home), the SLAM algorithm will perform a loop closure. This action will result in a pose adjustment and updates to the SLAM map.

However, this might lead to inconsistencies in the Exploration Signed Distance Field (ESDF) map used by the exploration planner, which might not be updated in line with the SLAM adjustments. What is the best approach for the planner to handle these pose jumps after a SLAM loop closure? Should the ESDF map be reconstructed from the updated SLAM map sections? Additionally, how should the planner re-evaluate the frontiers after such updates?