Mountain road Rosbag

[TokyoWarfare]

Hello I will soon share this rosbag. Total 14Km 7 up 7 down

VIDEO
https://youtu.be/3FCpqiim2XY

VIDEO 2 , usign GPU Offload, max scanner resolution
https://www.youtube.com/watch?v=KzrmtFyXs1g

ROSBAG (ROS2)
https://drive.google.com/drive/folders/1LRXavg0I-kpvwakc4kpEGwYxVrYWVJbF?usp=sharing

Video shows odometry at max density without problem. Mapping is thinned because I still run the docker image without gpu mem offload.

There is some drift but not terrible, Surprised there is not much vertical drift like it happens in highway runs.

Data was captured at an average 35-40 Km/h .

I was able to correct the drift using the loop tool but doing so manually is tiresome. So now I'm ready to test Scancontext.

Will share the bag when I make room in my cloud space

[TokyoWarfare]

I was able to very nicely align after:

• 2 manual loop closures in the offline viewer:
• optimize
• find overlapping submaps
• optimize.

data has aligned very very nicely

cross section shows an absolutelly perfect alignment

[TokyoWarfare]

The SLAM config
00_CONFIG_CURRENT.zip

It is certainly sparse, but it is a great start. I'll try to achieve the same in the highway SLAM.