When using evo_traj, the attitude Angle cannot be aligned. about evo

[MichaelGrupp]

Why did you say you didn't use euler's Angle?

Because it's not using them. Euler angles are only used for easy-to-interpret visualization in the plot of evo_traj. All angular metrics (--pose_relation angle_deg / angle_rad) calculate the rotation angle using the logarithmic map of the SO(3) rotation matrix, as shown in the formulas mentioned above. Or check what the code does in https://github.com/MichaelGrupp/evo/blob/master/evo/core/metrics.py , e.g.:

...
elif self.pose_relation == PoseRelation.rotation_angle_rad:
    self.error = np.array([abs(lie.so3_log(E_i[:3, :3])) for E_i in self.E])
...

[mengqingyu123]

This is the result of running.

[MichaelGrupp]

Looks like a duplicate of: #23

[mengqingyu123]

I see. Thank you for your help.

[chenwinki]

@mengqingyu123 have you solve this problem? could you give me a hint? thank you !

[mengqingyu123]

@chenwinki I consulted MichaelGrupp suggestion, absolute euler Angle error is not easy to understand and have limitations, so I use the evo_rpe get some indicators of euler Angle error, rather than directly compare the pitching Angle, course Angle and roll Angle.

[chenwinki]

@mengqingyu123 thank you~~~

[MichaelGrupp]

I use the evo_rpe get some indicators of euler Angle error

This doesn't use Euler angles. Check the formulas in notebooks/metrics.py_API_Documentation.ipynb.

[mengqingyu123]

@MichaelGrupp Uh... Sorry, I want to ask whether the command "evo_rpe tum MH05_groundtruth. TXT msckf_mh05.txt-a --pose_relation angle_deg-p" outputs some indicators of euler Angle. Why did you say you didn't use euler's Angle? I'm sorry to bother you with this question.

[mengqingyu123]

@MichaelGrupp I see. Thank you very much for your patience.