embodichain.lab.task_program.integrations.simulation

embodichain.lab.task_program.integrations.simulation#

Simulation bindings and factories for Task Program integrations.

These immutable declarations bind canonical Task Program scene and robot contracts to explicitly selected simulation entities and control parts.

`AntipodalGraspAffordanceBinding`	Build one antipodal grasp affordance from a selected rigid-object mesh.
`ContainerAffordanceBinding`	Declare one exact object target frame inside a container parent.
`ControlPartCommandPreset`	Named one-dimensional joint commands for one exact control part.
`ControlPartEndpointBinding`	Profile endpoint backed by one explicit robot control part.
`ControlPartResourceBinding`	Joint-backed robot resource containing control-part endpoints.
`RigidizedArticulationAntipodalGraspBinding`	Build a root-frame grasp mesh from one rigidized articulation link.
`SimulationArticulationBinding`	Explicit binding for one simulation articulation.
`SimulationArticulationLinkBinding`	Explicit canonical link backed by one native articulation link.
`SimulationRigidizedArticulationObjectBinding`	Bind one locked, floating articulation as a semantic object.
`SimulationRigidObjectBinding`	Explicit binding for one simulation rigid object.
`SimulationRobotSkillProfileBinding`	Build a profile from typed resources with strict native validation.
`SimulationSceneBinding`	Build one authoritative registry from explicit simulation bindings.
`SupportSurfaceAffordanceBinding`	Declare one exact object target frame on a support parent.

Classes:

`AntipodalGraspAffordanceBinding`	Build one antipodal grasp affordance from a selected rigid-object mesh.
`ContainerAffordanceBinding`	Declare one exact object target frame inside a container parent.
`ControlPartCommandPreset`	Named one-dimensional joint commands for one exact control part.
`ControlPartEndpointBinding`	Profile endpoint backed by one explicit robot control part.
`ControlPartResourceBinding`	Joint-backed robot resource containing control-part endpoints.
`RigidizedArticulationAntipodalGraspBinding`	Build a root-frame grasp mesh from one rigidized articulation link.
`SimulationArticulationBinding`	Explicit binding for one simulation articulation.
`SimulationArticulationLinkBinding`	Explicit canonical link backed by one native articulation link.
`SimulationRigidObjectBinding`	Explicit binding for one simulation rigid object.
`SimulationRigidizedArticulationObjectBinding`	Bind one locked, floating articulation as a semantic object.
`SimulationRobotSkillProfileBinding`	Build a profile from typed resources with strict native validation.
`SimulationSceneBinding`	Build one authoritative registry from explicit simulation bindings.
`SupportSurfaceAffordanceBinding`	Declare one exact object target frame on a support parent.

class embodichain.lab.task_program.integrations.simulation.AntipodalGraspAffordanceBinding[source]#

Build one antipodal grasp affordance from a selected rigid-object mesh.

internal_axis upgrades the payload to AxisAlignAffordance while preserving the same antipodal grasp capability. The axis is expressed in the rigid object’s local frame and can be reused by held-object skills such as Pour.

Methods:

__init__(entity_id, object_id, native_name, ...)

Attributes:

`aliases`
`entity_id`
`internal_axis`
`mesh_env_id`
`native_name`
`object_id`
`relative_pose`
`revision`

__init__(entity_id, object_id, native_name, revision, aliases=(), relative_pose=(1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0), mesh_env_id=0, internal_axis=None)#

aliases: tuple[str, ...]#

entity_id: str#

internal_axis: tuple[float, float, float] | None#

mesh_env_id: int#

native_name: str#

object_id: str#

relative_pose: tuple[float, ...]#

revision: str#

class embodichain.lab.task_program.integrations.simulation.ContainerAffordanceBinding[source]#

Declare one exact object target frame inside a container parent.

Parameters:

entity_id (str) – Canonical ID of the placement affordance.
parent_id (str) – Canonical object, articulation, or link parent ID.
native_name (str) – Stable native name of this target frame.
aliases (tuple[str, ...]) – Optional non-authoritative lookup aliases.
object_target_pose (tuple[float, ...]) – Desired object pose relative to the parent.
object_target_frame (str) – parent composes the parent transform; world_axes uses its position with world-oriented target axes.
minimum_confidence (float) – Minimum parent/affordance observation confidence.
is_default (bool) – Whether this is the parent’s default Place(inside=...) frame.

Methods:

__init__(entity_id, parent_id, native_name)

Attributes:

`aliases`
`entity_id`
`is_default`
`minimum_confidence`
`native_name`
`object_target_frame`
`object_target_pose`
`parent_id`

__init__(entity_id, parent_id, native_name, aliases=(), object_target_pose=(1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0), minimum_confidence=0.0, is_default=False, object_target_frame='parent')#

aliases: tuple[str, ...]#

entity_id: str#

is_default: bool#

minimum_confidence: float#

native_name: str#

object_target_frame: str#

object_target_pose: tuple[float, ...]#

parent_id: str#

class embodichain.lab.task_program.integrations.simulation.ControlPartCommandPreset[source]#

Named one-dimensional joint commands for one exact control part.

Methods:

`__init__`(preset_id, control_part, commands)
`build`(*, control_dof)	Build a command profile after validating the native control width.
`declare`()	Build a provider-free command profile from declared tuple widths.

Attributes:

`commands`
`control_part`
`preset_id`

__init__(preset_id, control_part, commands)#

build(*, control_dof)[source]#

Build a command profile after validating the native control width.

Return type:: ControlPartCommandProfile

commands: Mapping[str, tuple[float, ...]]#

control_part: str#

declare()[source]#

Build a provider-free command profile from declared tuple widths.

Return type:: ControlPartCommandProfile

preset_id: str#

class embodichain.lab.task_program.integrations.simulation.ControlPartEndpointBinding[source]#

Profile endpoint backed by one explicit robot control part.

Methods:

`__init__`(endpoint_id, control_part, capabilities)
`build`(robot)	Build a joint-backed endpoint after native control-part validation.
`declare`()	Return the endpoint contract without reading a robot.

Attributes:

`capabilities`
`command_preset`
`control_part`
`endpoint_id`

__init__(endpoint_id, control_part, capabilities, command_preset=None)#

build(robot)[source]#

Build a joint-backed endpoint after native control-part validation.

Return type:: ResourceEndpoint

capabilities: frozenset[str]#

command_preset: str | None#

control_part: str#

declare()[source]#

Return the endpoint contract without reading a robot.

Return type:: ResourceEndpoint

endpoint_id: str#

class embodichain.lab.task_program.integrations.simulation.ControlPartResourceBinding[source]#

Joint-backed robot resource containing control-part endpoints.

Methods:

`__init__`(resource_id[, endpoints, members])
`build`(robot)	Build a resource containing strictly validated control-part endpoints.
`declare`()	Return the resource graph without reading native control parts.

Attributes:

`endpoints`
`members`
`resource_id`

__init__(resource_id, endpoints=(), members=())#

build(robot)[source]#

Build a resource containing strictly validated control-part endpoints.

Return type:: RobotResource

declare()[source]#

Return the resource graph without reading native control parts.

Return type:: RobotResource

endpoints: tuple[ControlPartEndpointBinding, ...]#

members: tuple[str, ...]#

resource_id: str#

class embodichain.lab.task_program.integrations.simulation.RigidizedArticulationAntipodalGraspBinding[source]#

Build a root-frame grasp mesh from one rigidized articulation link.

Parameters:

entity_id (str) – Canonical affordance ID.
object_id (str) – Parent rigidized-articulation object ID.
grasp_link (str) – Native link supplying the grasp mesh.
native_name (str) – Stable native name for the affordance.
revision (str) – Affordance payload revision.
aliases (tuple[str, ...]) – Optional non-authoritative lookup aliases.
relative_pose (tuple[float, ...]) – Flattened affordance-to-object SE(3) transform.

Methods:

__init__(entity_id, object_id, grasp_link, ...)

Attributes:

`aliases`
`entity_id`
`grasp_link`
`native_name`
`object_id`
`relative_pose`
`revision`

__init__(entity_id, object_id, grasp_link, native_name, revision, aliases=(), relative_pose=(1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0))#

aliases: tuple[str, ...]#

entity_id: str#

grasp_link: str#

native_name: str#

object_id: str#

relative_pose: tuple[float, ...]#

revision: str#

class embodichain.lab.task_program.integrations.simulation.SimulationArticulationBinding[source]#

Explicit binding for one simulation articulation.

Methods:

__init__(entity_id, simulation_uid[, ...])

Attributes:

`aliases`
`collision_role`
`dynamics`
`entity_id`
`geometry_provider`
`semantic_type`
`simulation_uid`

__init__(entity_id, simulation_uid, aliases=(), dynamics=SceneDynamics.UNKNOWN, collision_role=SceneCollisionRole.NONE, semantic_type=None, geometry_provider=None)#

aliases: tuple[str, ...]#

collision_role: SceneCollisionRole#

dynamics: SceneDynamics#

entity_id: str#

geometry_provider: SceneGeometryProvider | None#

semantic_type: str | None#

simulation_uid: str#

class embodichain.lab.task_program.integrations.simulation.SimulationArticulationLinkBinding[source]#

Explicit canonical link backed by one native articulation link.

Methods:

__init__(entity_id, articulation_id, ...[, ...])

Attributes:

`aliases`
`articulation_id`
`dynamics`
`entity_id`
`native_link_name`
`semantic_type`

__init__(entity_id, articulation_id, native_link_name, aliases=(), dynamics=SceneDynamics.UNKNOWN, semantic_type=None)#

aliases: tuple[str, ...]#

articulation_id: str#

dynamics: SceneDynamics#

entity_id: str#

native_link_name: str#

semantic_type: str | None#

class embodichain.lab.task_program.integrations.simulation.SimulationRigidObjectBinding[source]#

Explicit binding for one simulation rigid object.

Methods:

__init__(entity_id, simulation_uid[, ...])

Attributes:

`aliases`
`collision_role`
`default_grasp_affordance`
`dynamics`
`entity_id`
`geometry_provider`
`semantic_type`
`simulation_uid`

__init__(entity_id, simulation_uid, aliases=(), dynamics=SceneDynamics.UNKNOWN, collision_role=SceneCollisionRole.NONE, semantic_type=None, default_grasp_affordance=None, geometry_provider=None)#

aliases: tuple[str, ...]#

collision_role: SceneCollisionRole#

default_grasp_affordance: str | None#

dynamics: SceneDynamics#

entity_id: str#

geometry_provider: SceneGeometryProvider | None#

semantic_type: str | None#

simulation_uid: str#

class embodichain.lab.task_program.integrations.simulation.SimulationRigidizedArticulationObjectBinding[source]#

Bind one locked, floating articulation as a semantic object.

Parameters:

entity_id (str) – Canonical scene object ID.
simulation_uid (str) – Explicit native articulation UID.
locked_qpos (Mapping[str, float]) – Exact locked position for every native joint.
aliases (tuple[str, ...]) – Optional non-authoritative lookup aliases.
dynamics (SceneDynamics) – Semantic dynamics classification.
collision_role (SceneCollisionRole) – Collision-world participation role.
semantic_type (str | None) – Optional semantic object classification.
default_grasp_affordance (str | None) – Optional default grasp affordance ID.
geometry_provider (SceneGeometryProvider | None) – Optional planner-facing collision geometry provider.
joint_position_tolerance (float) – Positive joint-state tolerance no greater than 1e-3. Configured limit coincidence uses a separate fixed epsilon.
link_transform_tolerance (float) – Positive cross-arena transform tolerance no greater than 1e-5.

Attention

Live configured assembly accepts only floating roots whose native joint limits are finite and coincident at every declared lock position. The coincidence proof uses a fixed safety epsilon and cannot be weakened by an authoring tolerance.

Methods:

__init__(entity_id, simulation_uid, locked_qpos)

Attributes:

`aliases`
`collision_role`
`default_grasp_affordance`
`dynamics`
`entity_id`
`geometry_provider`
`joint_position_tolerance`
`link_transform_tolerance`
`locked_qpos`
`semantic_type`
`simulation_uid`

__init__(entity_id, simulation_uid, locked_qpos, aliases=(), dynamics=SceneDynamics.UNKNOWN, collision_role=SceneCollisionRole.NONE, semantic_type=None, default_grasp_affordance=None, geometry_provider=None, joint_position_tolerance=0.001, link_transform_tolerance=1e-05)#

aliases: tuple[str, ...]#

collision_role: SceneCollisionRole#

default_grasp_affordance: str | None#

dynamics: SceneDynamics#

entity_id: str#

geometry_provider: SceneGeometryProvider | None#

joint_position_tolerance: float#

link_transform_tolerance: float#

locked_qpos: Mapping[str, float]#

semantic_type: str | None#

simulation_uid: str#

class embodichain.lab.task_program.integrations.simulation.SimulationRobotSkillProfileBinding[source]#

Build a profile from typed resources with strict native validation.

Methods:

`__init__`(profile_id, resources[, ...])
`build`(robot)	Build the existing profile after validating every typed resource.
`declare`()	Project the complete provider-free robot skill profile.

Attributes:

`command_presets`
`default_preset`
`defaults`
`grounding_providers`
`presets`
`profile_id`
`resources`
`skill_presets`

__init__(profile_id, resources, command_presets=(), defaults=<factory>, presets=(), default_preset=None, skill_presets=<factory>, grounding_providers=<factory>)#

build(robot)[source]#

Build the existing profile after validating every typed resource.

Parameters:: robot (Robot) – Live robot selected by the simulation factory.
Return type:: RobotSkillProfile
Returns:: Reusable, engine-independent robot skill profile.

command_presets: tuple[ControlPartCommandPreset, ...]#

declare()[source]#

Project the complete provider-free robot skill profile.

Return type:: RobotSkillProfile

default_preset: str | None#

defaults: Mapping[str, Mapping[str, str]]#

grounding_providers: Mapping[str, str]#

presets: tuple[SkillPolicyPreset, ...]#

profile_id: str#

resources: tuple[ControlPartResourceBinding | RobotResource, ...]#

skill_presets: Mapping[str, str]#

class embodichain.lab.task_program.integrations.simulation.SimulationSceneBinding[source]#

Build one authoritative registry from explicit simulation bindings.

Methods:

`__init__`(registry_id[, rigid_objects, ...])
`build`(simulation)	Build the existing authoritative scene registry.
`declare`()	Project the complete provider-free scene declaration.

Attributes:

`antipodal_grasps`
`articulations`
`collision_world_mode`
`containers`
`links`
`registry_id`
`rigid_objects`
`rigidized_articulation_grasps`
`rigidized_articulations`
`support_surfaces`

__init__(registry_id, rigid_objects=(), articulations=(), links=(), antipodal_grasps=(), support_surfaces=(), containers=(), collision_world_mode=None, *, rigidized_articulations=(), rigidized_articulation_grasps=())#

antipodal_grasps: tuple[AntipodalGraspAffordanceBinding, ...]#

articulations: tuple[SimulationArticulationBinding, ...]#

build(simulation)[source]#

Build the existing authoritative scene registry.

Parameters:: simulation (SimulationManager) – Live simulation used only for explicitly named lookups.
Return type:: SceneRegistry
Returns:: Immutable registry with typed roots, links, and affordances.

collision_world_mode: SceneCollisionWorldMode | None#

containers: tuple[ContainerAffordanceBinding, ...]#

declare()[source]#

Project the complete provider-free scene declaration.

Return type:: SceneManifest

links: tuple[SimulationArticulationLinkBinding, ...]#

registry_id: str#

rigid_objects: tuple[SimulationRigidObjectBinding, ...]#

rigidized_articulation_grasps: tuple[RigidizedArticulationAntipodalGraspBinding, ...]#

rigidized_articulations: tuple[SimulationRigidizedArticulationObjectBinding, ...]#

support_surfaces: tuple[SupportSurfaceAffordanceBinding, ...]#

class embodichain.lab.task_program.integrations.simulation.SupportSurfaceAffordanceBinding[source]#

Declare one exact object target frame on a support parent.

Parameters:

entity_id (str) – Canonical ID of the placement affordance.
parent_id (str) – Canonical object, articulation, or link parent ID.
native_name (str) – Stable native name of this target frame.
aliases (tuple[str, ...]) – Optional non-authoritative lookup aliases.
object_target_pose (tuple[float, ...]) – Desired object pose relative to the parent.
object_target_frame (str) – parent composes the complete parent transform; world_axes adds its translation to the parent position while keeping the declared orientation and offset in world axes.
minimum_confidence (float) – Minimum parent/affordance observation confidence.
is_default (bool) – Whether this is the parent’s default Place(on=...) frame.

Methods:

__init__(entity_id, parent_id, native_name)

Attributes:

`aliases`
`entity_id`
`is_default`
`minimum_confidence`
`native_name`
`object_target_frame`
`object_target_pose`
`parent_id`

__init__(entity_id, parent_id, native_name, aliases=(), object_target_pose=(1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 0.0, 1.0), minimum_confidence=0.0, is_default=False, object_target_frame='parent')#

aliases: tuple[str, ...]#

entity_id: str#

is_default: bool#

minimum_confidence: float#

native_name: str#

object_target_frame: str#

object_target_pose: tuple[float, ...]#

parent_id: str#

Scene preparation and workspace#

Scene proposal contracts remain in lab.sim.scene.expansion. These adapters connect them to the current robot and the Gym reset lifecycle:

embodichain.lab.task_program.integrations.simulation

Contents

embodichain.lab.task_program.integrations.simulation#

Scene preparation and workspace#