Closed marleyshan21 closed 8 months ago
@marleyshan21 I'd rather load the output gtiff into apps like QGIS and check there. I am pretty sure some of the static transformers is publishing a wrong transformation or the satellite map uses a wrong transformation. The output data is unaffected by that, just a visualisation issue. Not sure if I find the time fix it.
Ah okay ✌️! I will try it out on QGIS.
Also, I will work on checking this static transform part. If you could guide me on where this static transform part exists in the code, i could check it out and try fixing it.
Lemme know!
I tried it with QGIS. The results are a lot better with slight differences w.r.t the ground truth
I will checkout the code regarding this static transform part. I could use your help in zeroing in on the file and function where these things happen.
If it's on OpenREALM side it is somewhere here:
Have you tried the rviz_satellite fork from my GitHub?
I think I remember there was an issue with some frames being redefined that caused the misalignment.
Using your fork of rviz_satellite solved it! https://github.com/laxnpander/rviz_satellite
Based on your commit, it looks like an issue with the frame being lookedup to determine the transformation. Your fork uses the one needed by OpenRealm
Thanks for the help @laxnpander!
Sure, glad I could help.
Hi!
I was running the test dataset in 2D mapping with visual SLAM mode. I am seeing disparity in the results when compared to the video of OpenRealm.
I tried playing around primarily with nrof_features for the orb extractor and ground sampling distance. What are the other main parameters that i need to tune? The images are for the default parameters that are present in the repo.
@laxnpander