Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

The metrics evaluation is not working #462

Open
setareh-soltanieh opened this issue Aug 9, 2024 · 2 comments
Open

The metrics evaluation is not working #462

setareh-soltanieh opened this issue Aug 9, 2024 · 2 comments
Labels
question Theory or implementation question

Comments

@setareh-soltanieh
Copy link

setareh-soltanieh commented Aug 9, 2024

I want to compare the estimated pose with the ground truth pose using ov_eval. The dataset that I am using is the EuRoC dataset. I am recording a rosbag of the estimated pose data (/ov_msckf/poseimu) and the ground truth poses (for the vicon_room1 is: /vicon/firefly_sbx/firefly_sbx). After that I convert this rosbag into two txt files (stamped_traj_estimate.txt and stamped_groundtruth.txt). After that I run the following command to plot my results.

rosrun ov_eval plot_trajectories stamped_groundtruth.txt stamped_traj_estimate.txt
Figure_1
Figure_2

This command works and it will plot the ground truth and the aligned estimated paths correctly. but when I want to capture the ATE and RPE errors using the following command:

rosrun ov_eval error_singlerun stamped_groundtruth.txt stamped_traj_estimate.txt

The results don't make sense!

======================================
Absolute Trajectory Error
======================================
rmse_ori = 178.702 | rmse_pos = 0.103
mean_ori = 178.701 | mean_pos = 0.097
min_ori  = 176.714 | min_pos  = 0.016
max_ori  = 179.984 | max_pos  = 0.173
std_ori  = 0.547 | std_pos  = 0.037
======================================
Relative Pose Error
======================================
seg 8 - median_ori = 143.450 | median_pos = 4.591 (2383 samples)
seg 16 - median_ori = 88.770 | median_pos = 3.702 (2082 samples)
seg 24 - median_ori = 92.984 | median_pos = 3.481 (1816 samples)
seg 32 - median_ori = 85.973 | median_pos = 3.481 (1427 samples)
seg 40 - median_ori = 138.841 | median_pos = 3.082 (1099 samples)

The orientations are almost 180 degrees incorrect. Can someone please tell me what is the issue?

@goldbattle
Copy link
Member

Hi, you shouldn't directly use the /vicon/firefly_sbx/firefly_sbx topic since this contains the transform from the mocap world frame to the mocap marker frame. This isn't the IMU sensor frame we estimate in. You can either download the tar files and get the processed groundtruth files from the authors, or use the ones we provide in the repo:
https://github.com/rpng/open_vins/tree/master/ov_data/euroc_mav

Hope this helps.

@goldbattle goldbattle added the question Theory or implementation question label Aug 12, 2024
@setareh-soltanieh
Copy link
Author

Hi @goldbattle,
It worked. Thanks a lot!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Theory or implementation question
Projects
None yet
Development

No branches or pull requests

2 participants