def reorder_actions()

in DownstreamRL/TrainZPolicyRL.py [0:0]

• 5 lines of code
• 10 McCabe index (conditional complexity)

	def reorder_actions(self, actions):

		# Assume that the actions are 16 dimensional, and are ordered as: 
		# 7 DoF for left arm, 7 DoF for right arm, 1 for left gripper, and 1 for right gripper. 

		# The original trajectory has gripper values from 0 (Close) to 1 (Open), but we've to rescale to -1 (Open) to 1 (Close) for Mujoco. 
		# And handle joint velocities.
		# MIME Gripper values are from 0 to 100 (Close to Open), but we assume actions has values from 0 to 1 (Close to Open), and then rescale to (-1 Open to 1 Close) for Mujoco.
		# Mujoco needs them flipped.

		indices = np.array([ 7,  8,  9, 10, 11, 12, 13,  0,  1,  2,  3,  4,  5,  6, 15, 14])
		reordered_actions = actions[:,indices]
		reordered_actions[:,14:] = 1 - 2*reordered_actions[:,14:]
		return reordered_actions