Case of noisy data

[MariaMel98]

Hello, I wanted to ask if the code described in rigid_transform_3D is working in case of noisy data.
I created a random point cloud (30 points) which is A and then transformed it like B = R*A + t.
Then I created some noise with mean = 0 and std=0.1 with: noise = np.random.normal(0, noise_std, B.shape) and I added it to B, so B_noised = B+noise. After testing the code on the set (A, B_noised) I got an rmse of 0.03 which is not that bad.
In case of real world 3D data with possibly random noise will this code work ?
Thanks in advance!

[nghiaho12]

If the noise is zero mean and Gaussian it should work The algorithm solves a least square problem.