jzhuai0108/fast_lio_slam

LiDAR SLAM = FAST-LIO + Scan Context

fast_lio_slam's Issues

Performance of the resulting maps (integrated with PGO) are not significantly improved

Dear Dr. Jianzhu(Josh) Huai,

I have run the package in the terminal and closed loop detected is printed in the terminal during runtime, However, the resulting maps from Fast-LIO + PGO and Fast-LIO alone appear to have the same performance as shown in the first figure. I've checked the source code at commit e3f6278, and all the parameters related to loop closure seem to be properly enabled, including the use of scan-context. Could you please help me understand why the integration of PGO isn't significantly improving map quality as expected? Are there any specific troubleshooting steps or configurations I should consider to address this issue? Right now my robot using velodyne sensor.

Last night, I tried your latest version of the source code (44a17c1) which performs the radius search algorithm. However, the resulting maps I've generated fall short of my expectations as shown in the second figure. While I believe your code is a valuable resource, it's possible that the parameters I've configured need adjustment? Your insights would be greatly appreciated, and the attached maps should provide a clear picture of the challenges I'm facing.

Thank you for your assistance.