rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils

1.  Module Contents

1.1.  Functions

compute_heading_and_up
compute_rot
quat_axis
scale_transform	Normalizes a given input tensor to a range of [-1, 1].
unscale_transform	Denormalizes a given input tensor from range of [-1, 1] to (lower, upper).
saturate	Clamps a given input tensor to (lower, upper).
quat_diff_rad	Get the difference in radians between two quaternions.
local_to_world_space	Convert a point from the local frame to the global frame Args:
normalise_quat_in_pose	Takes a pose and normalises the quaternion portion of it.
my_quat_rotate
quat_to_angle_axis
angle_axis_to_exp_map
quat_to_exp_map
quat_to_tan_norm
euler_xyz_to_exp_map
exp_map_to_angle_axis
exp_map_to_quat
slerp
calc_heading
calc_heading_quat
calc_heading_quat_inv

1.2.  API

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.compute_heading_and_up(torso_rotation: Tensor, inv_start_rot: Tensor, to_target: Tensor, vec0: Tensor, vec1: Tensor, up_idx: int) → Tuple[Tensor, Tensor, Tensor, Tensor, Tensor]

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.compute_rot(torso_quat, velocity, ang_velocity, targets, torso_positions)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.quat_axis(q: Tensor, axis: int = 0) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.scale_transform(x: isaacgym.torch_utils.torch.Tensor, lower: isaacgym.torch_utils.torch.Tensor, upper: isaacgym.torch_utils.torch.Tensor) → isaacgym.torch_utils.torch.Tensor

Normalizes a given input tensor to a range of [-1, 1].

@note It uses pytorch broadcasting functionality to deal with batched input.

Args:

x: Input tensor of shape (N, dims). lower: The minimum value of the tensor. Shape (dims,) upper: The maximum value of the tensor. Shape (dims,)

Returns:

Normalized transform of the tensor. Shape (N, dims)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.unscale_transform(x: isaacgym.torch_utils.torch.Tensor, lower: isaacgym.torch_utils.torch.Tensor, upper: isaacgym.torch_utils.torch.Tensor) → isaacgym.torch_utils.torch.Tensor

Denormalizes a given input tensor from range of [-1, 1] to (lower, upper).

@note It uses pytorch broadcasting functionality to deal with batched input.

Args:

x: Input tensor of shape (N, dims). lower: The minimum value of the tensor. Shape (dims,) upper: The maximum value of the tensor. Shape (dims,)

Returns:

Denormalized transform of the tensor. Shape (N, dims)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.saturate(x: isaacgym.torch_utils.torch.Tensor, lower: isaacgym.torch_utils.torch.Tensor, upper: isaacgym.torch_utils.torch.Tensor) → isaacgym.torch_utils.torch.Tensor

Clamps a given input tensor to (lower, upper).

@note It uses pytorch broadcasting functionality to deal with batched input.

Args:

x: Input tensor of shape (N, dims). lower: The minimum value of the tensor. Shape (dims,) upper: The maximum value of the tensor. Shape (dims,)

Returns:

Clamped transform of the tensor. Shape (N, dims)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.quat_diff_rad(a: isaacgym.torch_utils.torch.Tensor, b: isaacgym.torch_utils.torch.Tensor) → isaacgym.torch_utils.torch.Tensor

Get the difference in radians between two quaternions.

Args:

a: first quaternion, shape (N, 4) b: second quaternion, shape (N, 4)

Returns:

Difference in radians, shape (N,)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.local_to_world_space(pos_offset_local: isaacgym.torch_utils.torch.Tensor, pose_global: isaacgym.torch_utils.torch.Tensor)

Convert a point from the local frame to the global frame Args:

pos_offset_local: Point in local frame. Shape: [N, 3] pose_global: The spatial pose of this point. Shape: [N, 7]

Returns:

Position in the global frame. Shape: [N, 3]

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.normalise_quat_in_pose(pose)

Takes a pose and normalises the quaternion portion of it.

Args:

pose: shape N, 7

Returns:

Pose with normalised quat. Shape N, 7

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.my_quat_rotate(q, v)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.quat_to_angle_axis(q: Tensor) → Tuple[Tensor, Tensor]

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.angle_axis_to_exp_map(angle: Tensor, axis: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.quat_to_exp_map(q: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.quat_to_tan_norm(q: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.euler_xyz_to_exp_map(roll: Tensor, pitch: Tensor, yaw: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.exp_map_to_angle_axis(exp_map)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.exp_map_to_quat(exp_map)

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.slerp(q0: Tensor, q1: Tensor, t: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.calc_heading(q: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.calc_heading_quat(q: Tensor) → Tensor

rofunc.learning.RofuncRL.tasks.utils.torch_jit_utils.calc_heading_quat_inv(q: Tensor) → Tensor