The measurement method of the rotation angle may be wrong

[Theigrams]

https://github.com/XiaoshuiHuang/CSBaselineAlgo/blob/0f15dd81052dcfd073a9024f65365ad16e48442e/Baselines/JRMPC/RunJPMPCT_cs.m#LL92C1-L98C51

RR=R{1}\R{2};
angle_est = rotm2eul(RR);
angle_mse=norm(angle_est-angle_gt,'fro')*180/3.14;

There use the norm of difference between the axis-angle vectors $||r_1-r_2||_2$ , which seems to be the difference angle of two rotation.

However, we have

which because

We need to use the Baker–Campbell–Hausdorff formula to calculate it.