all paths start from the same origin in rviz

[BamsaAhmed]

when apply them in rivz all paths start from the same point, how should run them and keeping each robot from it's own start point?

[yunzc]

Hi. Are you visualizing the trajectory of Kimera-VIO or Kimera-Multi?

[BamsaAhmed]

Hi I used kimera_multi dataset to implement MR-SLAM, but i don't know the starting coordinates to start running the dataset so i start all at origin coordinate (0,0,0) at rviz, this problem causes no overlap area between the trajectories or many wrong overlapped area. The final constructed map is not matched with the groundtruth map, so how can i resolve this problem?

‫في الاثنين، 1 مايو 2023 في 5:14 م تمت كتابة ما يلي بواسطة ‪Yun Chang‬‏ <‪ @.***‬‏>:‬

Hi. Are you visualizing the trajectory of Kimera-VIO or Kimera-Multi?

— Reply to this email directly, view it on GitHub https://github.com/MIT-SPARK/Kimera-Multi-Data/issues/4#issuecomment-1529759379, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZHQ4Z6A4V73R4V3VQVDCI3XD7ANNANCNFSM6AAAAAAXRSCNSE . You are receiving this because you authored the thread.Message ID: @.***>

[yunzc]

You can align the trajectories with the ground truth trajectories given and use the resulting transform to transform your constructed map.

[BamsaAhmed]

How can we do it? Can you support me with any code or guide?

‫في الثلاثاء، 9 مايو 2023 في 7:02 م تمت كتابة ما يلي بواسطة ‪Yun Chang‬‏ <‪ @.***‬‏>:‬

You can align the trajectories with the ground truth trajectories given and use the resulting transform to transform your constructed map.

— Reply to this email directly, view it on GitHub https://github.com/MIT-SPARK/Kimera-Multi-Data/issues/4#issuecomment-1540462438, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZHQ4Z3SY5RTY2RYS5MHVDLXFJTC5ANCNFSM6AAAAAAXRSCNSE . You are receiving this because you authored the thread.Message ID: @.***>

[yunzc]

Apologies, I just re-read your question:

Hi I used kimera_multi dataset to implement MR-SLAM, but i don't know the starting coordinates to start running the dataset so i start all at origin coordinate (0,0,0) at rviz, this problem causes no overlap area between the trajectories or many wrong overlapped area. The final constructed map is not matched with the groundtruth map, so how can i resolve this problem? ‫

I think this is more of a fundamental problem with the MR-SLAM algorithm. I would recommend you to check out our multi-robot SLAM system, Kimera-Multi. We are able to deal with scenarios without initial calibrations on locations. https://github.com/MIT-SPARK/Kimera-Multi

Closing this thread now because the issue is irrelevant to the datasets.