This project implements rotating a point cloud by randomly generating a rotation angle and unit vector.
I have implemented a rotation matrix and a quaternion, respectively, and visualized the results.

Given a rotation angle $\phi$ and a rotation vector (unit vector) $\vec{u}$ as input,
the following formulas can be utilized to implement a rotation matrix and a quaternion, respectively.

Given input point cloud of $x$, applying the rotation operator to each of them gives the following representation.

$$ R = exp([ \phi \cdot \vec{u} ]_{\times}) $$

$$ x' = R x $$

$$ q = exp( \phi \cdot \vec{u} / 2) $$

$$ x' = q \otimes x \otimes q^* $$

Refer to requirements.txt to setup the dependencies and execute the code like below.
I tested this on ubuntu 18.04.

python main.py --input_path ./bunny.ply

• Sola, Joan. "Quaternion kinematics for the error-state Kalman filter." arXiv preprint arXiv:1711.02508 (2017).
• Tricks to do annotation : isl-org/Open3D#2