3D plot from SLAM pose estimations

44 views Asked by At

I am working with a SLAM system that is based on ORB-SLAM3. I am at the results stage, where I am trying to write a python script to take the SLAM output files trajectory.txt and trajectory_keyframes.txt and make a 3d plot showing how the camera has moved through space.

Now, the data in both .txt files is of the following format:

"1705678100257923072.000000 -0.058561411 0.003284361 0.045478430 0.104412355 0.133441061 0.015586698 0.985418022
1705678100291521280.000000 -0.059866361 0.002477865 0.045944497 0.104993664 0.134505078 0.014572186 0.985227108
1705678100308320256.000000 -0.060874972 0.002395819 0.046614237 0.105552807 0.135332182 0.013078596 0.985075057
1705678100341918720.000000 -0.064644001 0.002536844 0.047003917 0.104946494 0.139216498 0.010510575 0.984629214
..." which presumably corresponds to timestamp, translations in x y z, and quaternions

.

So far I have been taking the translation values and plot them in a x-y-z graph. However, not only start and end point are far away from each other (which shouldn't be the case given that I started and stopped the recording in the same place) and I can't really trace back my movements either.

What step of the process am I missing?

0

There are 0 answers