[Question] How are the .npy and .pickle files that contain the grasping poses structured?

[Polterino]

Hi, I successfully generated some grasping poses of a model but I want to rotate and translate them. I have the rotation matrix and the translation vector, so I tried something like this:

import pickle
import numpy as np

poses_npy = np.load('003_cracker_box_2.npy')

transformation_matrix = np.eye(4)
rotation_matrix = np.array([[0.866, -0.5, 0, 0],
                            [0.5, 0.866, 0, 0],
                            [0, 0, 1, 0],
                            [0, 0, 0, 1]])
translation_vector = np.array([0.1, 0.2, 0.05, 1])
transformation_matrix[:3, :3] = rotation_matrix[:3, :3]
transformation_matrix[:3, 3] = translation_vector[:3]

transformed_poses_npy = []
for pose in poses_npy:
    pose_matrix = np.array(pose).reshape(3, 4)  # Assuming that it's a 1D array of length 12
    pose_matrix = np.vstack([pose_matrix, [0, 0, 0, 1]])  # Adding [0, 0, 0, 1] to the last row
    transformed_pose = np.dot(transformation_matrix, pose_matrix)
    transformed_poses_npy.append(transformed_pose)

np.save('transformed_poses.npy', transformed_poses_npy)

The program compiles, but when I try to display the transformed poses using read_grasps_from_file.py I get this error:

IndexError: index 6 is out of bounds for axis 0 with size 4

First of all, I'm not sure if I need to transorm both the .npy and the .pickle files or only the .npy.
And then I don't understand how the data in these files is structured, am I supposed to use a 3x3 rotation matrix and a 3D translation vector or a 4x4 rotation matrix and a 4D translation vector like in the code above?