This pull request fixes two small issues with Eigen::aligned_allocator so it can be build on Ubuntu 20.04.
Furthermore I added a script (evaluation/tum-vi_run_all.py) that runs many sequences of the TUM-VI dataset multiple times and reports the median RMS ATE. For this the evaluation script from TUM-RGBD (modified by ORB-SLAM3) is used.
I get the following results:
sequence name: median RMS ATE in m, fail count/ runs per sequence
room1: 0.087204, 0/3
room2: 0.109074, 0/3
room3: 0.110926, 0/3
room4: 0.076465, 0/3
room5: 0.095403, 2/3
room6: 0.082724, 0/3
This pull request fixes two small issues with
Eigen::aligned_allocator
so it can be build on Ubuntu 20.04.Furthermore I added a script (
evaluation/tum-vi_run_all.py
) that runs many sequences of the TUM-VI dataset multiple times and reports the median RMS ATE. For this the evaluation script from TUM-RGBD (modified by ORB-SLAM3) is used.I get the following results:
sequence name: median RMS ATE in m, fail count/ runs per sequence room1: 0.087204, 0/3 room2: 0.109074, 0/3 room3: 0.110926, 0/3 room4: 0.076465, 0/3 room5: 0.095403, 2/3 room6: 0.082724, 0/3
Room 1: