foundation-policy 0.0.9

A foundation model for continuous control

A Foundation Model for Continuous Control

Supports Python >= 3.6

from foundation_model import Quadrotor
import numpy as np
from scipy.spatial.transform import Rotation as R

quadrotor = Quadrotor()
position = [0, 0, 0]
orientation = [1, 0, 0, 0]
orientation_rotation_matrix = R.from_quat(np.array(orientation)[[1, 2, 3, 0]]).as_matrix().ravel().tolist()
linear_velocity = [0, 0, 0]
angular_velocity = [0, 0, 0]
previous_action = [0, 0, 0, 0]

observation = np.array([position + orientation_rotation_matrix + linear_velocity + angular_velocity + previous_action])
quadrotor.reset()
action = quadrotor.evaluate_step(observation)
print(action)