mani_skill.utils.sapien_utils#

Utilities that work with the simulation / SAPIEN

Attributes#

T

Functions#

apply_urdf_config(loader, urdf_config)

check_actor_static(actor[, lin_thresh, ang_thresh])

check_joint_stuck(articulation, active_joint_idx[, ...])

check_urdf_config(urdf_config)

Check whether the urdf config is valid for SAPIEN.

compute_total_impulse(contact_infos)

create_viewer(viewer_camera_config)

Creates a viewer with the given camera config

get_actor_state(actor)

get_articulation_padded_state(articulation, max_dof)

get_articulation_state(articulation)

get_cpu_actor_contacts(contacts, actor)

get_cpu_actors_contacts(contacts, actors)

This function is used to avoid double for-loop when using get_actor_contacts with multiple actors

get_multiple_pairwise_contacts(contacts, actor0, ...)

Given a list of contacts, return the dict of contacts involving the one actor and actors

get_obj_by_name(objs, name[, is_unique])

Get a object given the name.

get_obj_by_type(objs, target_type[, is_unique])

get_objs_by_names(objs, names)

Get a list of objects given a list of names from a larger list of objects (objs). The returned list is in the order of the names given

get_pairwise_contact_impulse(contacts, actor0, actor1)

get_pairwise_contacts(contacts, actor0, actor1)

Given a list of contacts, return the list of contacts involving the two actors

hex2rgba(h[, correction])

is_state_dict_consistent(state_dict)

Checks if the given state dictionary (generated via env.get_state_dict()) is consistent where each actor/articulation has the same batch dimension

look_at(eye, target[, up, device])

Get the camera pose in SAPIEN by the Look-At method.

parse_urdf_config(config_dict)

Parse config from dict for SAPIEN URDF loader.

sapien_pose_to_opencv_extrinsic(sapien_pose_matrix)

set_articulation_render_material(articulation, **kwargs)

set_render_material(material, **kwargs)

Module Contents#

mani_skill.utils.sapien_utils.apply_urdf_config(loader, urdf_config)[source]#
Parameters:

  • loader (sapien.wrapper.urdf_loader.URDFLoader) –

  • urdf_config (dict) –

mani_skill.utils.sapien_utils.check_actor_static(actor, lin_thresh=0.001, ang_thresh=0.01)[source]#
Parameters:

actor (mani_skill.utils.structs.actor.Actor) –

mani_skill.utils.sapien_utils.check_joint_stuck(articulation, active_joint_idx, pos_diff_threshold=0.001, vel_threshold=0.0001)[source]#
Parameters:

  • articulation (sapien.physx.PhysxArticulation) –

  • active_joint_idx (int) –

  • pos_diff_threshold (float) –

  • vel_threshold (float) –

mani_skill.utils.sapien_utils.check_urdf_config(urdf_config)[source]#

Check whether the urdf config is valid for SAPIEN.

Parameters:

urdf_config (dict) – dict passed to sapien.URDFLoader.load.

mani_skill.utils.sapien_utils.compute_total_impulse(contact_infos)[source]#
Parameters:

contact_infos (list[Tuple[sapien.physx.PhysxContact, bool]]) –

mani_skill.utils.sapien_utils.create_viewer(viewer_camera_config)[source]#

Creates a viewer with the given camera config

Parameters:

viewer_camera_config (mani_skill.sensors.camera.CameraConfig) –

mani_skill.utils.sapien_utils.get_actor_state(actor)[source]#
Parameters:

actor (sapien.Entity) –

mani_skill.utils.sapien_utils.get_articulation_padded_state(articulation, max_dof)[source]#
Parameters:

  • articulation (sapien.physx.PhysxArticulation) –

  • max_dof (int) –

mani_skill.utils.sapien_utils.get_articulation_state(articulation)[source]#
Parameters:

articulation (sapien.physx.PhysxArticulation) –

mani_skill.utils.sapien_utils.get_cpu_actor_contacts(contacts, actor)[source]#
Parameters:

  • contacts (list[sapien.physx.PhysxContact]) –

  • actor (sapien.Entity) –

Return type:

list[Tuple[sapien.physx.PhysxContact, bool]]

mani_skill.utils.sapien_utils.get_cpu_actors_contacts(contacts, actors)[source]#

This function is used to avoid double for-loop when using get_actor_contacts with multiple actors

Parameters:

  • contacts (list[sapien.physx.PhysxContact]) –

  • actors (list[sapien.Entity]) –

Return type:

dict[sapien.Entity, list[Tuple[sapien.physx.PhysxContact, bool]]]

mani_skill.utils.sapien_utils.get_multiple_pairwise_contacts(contacts, actor0, actor1_list)[source]#

Given a list of contacts, return the dict of contacts involving the one actor and actors This function is used to avoid double for-loop when using get_pairwise_contacts with multiple actors

Parameters:

  • contacts (list[sapien.physx.PhysxContact]) –

  • actor0 (sapien.Entity) –

  • actor1_list (list[sapien.Entity]) –

Return type:

dict[sapien.Entity, list[Tuple[sapien.physx.PhysxContact, bool]]]

mani_skill.utils.sapien_utils.get_obj_by_name(objs, name, is_unique=True)[source]#

Get a object given the name.

Parameters:

  • objs (list[T]) – objs to query. Expect these objects to have a get_name function. These may be sapien.Entity, physx.PhysxArticulationLink etc.

  • name (str) – name for query.

  • is_unique (bool, optional) – whether the name should be unique. Defaults to True.

Raises:

RuntimeError – The name is not unique when @is_unique is True.

Returns:

matched T or Ts. None if no matches.

Return type:

T or list[T]

mani_skill.utils.sapien_utils.get_obj_by_type(objs, target_type, is_unique=True)[source]#
Parameters:

  • objs (list[T]) –

  • target_type (T) –

mani_skill.utils.sapien_utils.get_objs_by_names(objs, names)[source]#

Get a list of objects given a list of names from a larger list of objects (objs). The returned list is in the order of the names given

Parameters:

  • objs (list[T]) – objs to query. Expect these objects to have a get_name function. These may be sapien.Entity, physx.PhysxArticulationLink etc.

  • name (str) – names to query.

  • names (list[str]) –

Returns:

matched T or Ts. None if no matches.

Return type:

T or list[T]

mani_skill.utils.sapien_utils.get_pairwise_contact_impulse(contacts, actor0, actor1)[source]#
Parameters:

  • contacts (list[sapien.physx.PhysxContact]) –

  • actor0 (sapien.Entity) –

  • actor1 (sapien.Entity) –

mani_skill.utils.sapien_utils.get_pairwise_contacts(contacts, actor0, actor1)[source]#

Given a list of contacts, return the list of contacts involving the two actors

Parameters:

  • contacts (list[sapien.physx.PhysxContact]) –

  • actor0 (sapien.Entity) –

  • actor1 (sapien.Entity) –

Return type:

list[Tuple[sapien.physx.PhysxContact, bool]]

mani_skill.utils.sapien_utils.hex2rgba(h, correction=True)[source]#
mani_skill.utils.sapien_utils.is_state_dict_consistent(state_dict)[source]#

Checks if the given state dictionary (generated via env.get_state_dict()) is consistent where each actor/articulation has the same batch dimension

Parameters:

state_dict (dict) –

mani_skill.utils.sapien_utils.look_at(eye, target, up=(0, 0, 1), device=None)[source]#

Get the camera pose in SAPIEN by the Look-At method.

Note

https://www.scratchapixel.com/lessons/mathematics-physics-for-computer-graphics/lookat-function The SAPIEN camera follows the convention: (forward, right, up) = (x, -y, z) while the OpenGL camera follows (forward, right, up) = (-z, x, y) Note that the camera coordinate system (OpenGL) is left-hand.

Parameters:

  • eye – camera location

  • target – looking-at location

  • up – a general direction of “up” from the camera.

  • device – device to put the pose on.

Returns:

camera pose

Return type:

Pose

mani_skill.utils.sapien_utils.parse_urdf_config(config_dict)[source]#

Parse config from dict for SAPIEN URDF loader.

Parameters:

config_dict (dict) – a dict containing link physical properties.

Returns:

urdf config passed to sapien.URDFLoader.load.

Return type:

dict

mani_skill.utils.sapien_utils.sapien_pose_to_opencv_extrinsic(sapien_pose_matrix)[source]#
Parameters:

sapien_pose_matrix (numpy.ndarray) –

Return type:

numpy.ndarray

mani_skill.utils.sapien_utils.set_articulation_render_material(articulation, **kwargs)[source]#
Parameters:

articulation (sapien.physx.PhysxArticulation) –

mani_skill.utils.sapien_utils.set_render_material(material, **kwargs)[source]#
Parameters:

material (sapien.render.RenderMaterial) –

mani_skill.utils.sapien_utils.T[source]#