The ATE would be much bigger

[FookSong]

Hey，Dr.Urban I tried to restore the example（indoor_dynamic）in your paper. I also try something else in LaFiDa ,It's a nice dataset. But unfortunately，I always got ATE 10 times(like the .jpg below ) than you wrote in your paper.I transformed my trajectory to the rigidbody ，but always the same.Is there anyway to get a better result of ATE. I think your work is really cool!!! So I want to restore it and do some experiments with MultiCol-SLAM. Any help(new version or something else) will be greatful!!

sincerely!