Improving ATE against ground truth using FAST_LIO_LOC with SLAM-generated maps

[Breaking-Bugs]

Thanks for providing this awesome repo! :)
I'm evaluating FAST_LIO_LOC for improving localization accuracy relative to ground truth. Here's my specific scenario:

1. I have a rosbag with accompanying ground truth trajectory data
2. When I run FAST_LIO2 (or another SLAM algorithm) on this rosbag, I get an ATE of approximately 1m compared to ground truth
3. I've generated a point cloud map using this SLAM run

My question is: Would running FAST_LIO_LOC on the same rosbag, using the map generated from the SLAM algorithm, likely reduce the ATE compared to ground truth? Or is the achievable accuracy fundamentally limited by the quality of the map that was generated with 1m ATE?

Any insights would be appreciated as I am quite new to this area.

[kainozoic]

Thanks for providing this awesome repo! :) I'm evaluating FAST_LIO_LOC for improving localization accuracy relative to ground truth. Here's my specific scenario:

1. I have a rosbag with accompanying ground truth trajectory data
2. When I run FAST_LIO2 (or another SLAM algorithm) on this rosbag, I get an ATE of approximately 1m compared to ground truth
3. I've generated a point cloud map using this SLAM run

My question is: Would running FAST_LIO_LOC on the same rosbag, using the map generated from the SLAM algorithm, likely reduce the ATE compared to ground truth? Or is the achievable accuracy fundamentally limited by the quality of the map that was generated with 1m ATE?

Any insights would be appreciated as I am quite new to this area.

Hello, I am currently studying in this area as well, and I am very interested in these issues. However, I am a beginner and know very little about it. May I ask if you have found the answers to these questions so far? Thank you!