MuJoCo Visualizer

MuJoCo Visualizer#

JAXsim provides a simple interface with MuJoCo’s visualizer. The visualizer is a separate process that communicates with the main simulation process. This allows for the simulation to run at full speed while the visualizer can run at a different frame rate.

Loaders#

class jaxsim.mujoco.loaders.MujocoCamera(mode='fixed', target=None, fovy='45', pos='0 0 0', quat=None, axisangle=None, xyaxes=None, zaxis=None, euler=None, name=None)[source]#

Helper class storing parameters of a Mujoco camera.

Refer to the official documentation for more details: https://mujoco.readthedocs.io/en/stable/XMLreference.html#body-camera

Parameters:

  • mode (str)

  • target (str | None)

  • fovy (str)

  • pos (str)

  • quat (str | None)

  • axisangle (str | None)

  • xyaxes (str | None)

  • zaxis (str | None)

  • euler (str | None)

  • name (str | None)

static build_from_target_view(camera_name, lookat=(0, 0, 0), distance=3, azimut=90, elevation=-45, fovy=45, degrees=True, **kwargs)[source]#

Create a custom camera that looks at a target point.

Note

The choice of the parameters is easier if we imagine to consider a target frame T whose origin is located over the lookat point and having the same orientation of the world frame W. We also introduce a camera frame C whose origin is located over the lower-left corner of the image, and having the x-axis pointing right and the y-axis pointing up in image coordinates. The camera renders what it sees in the -z direction of frame C.

Parameters:

  • camera_name (str) – The name of the camera.

  • lookat (Sequence[float | int] | ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]) – The target point to look at (origin of T).

  • distance (float | int | ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]) – The distance from the target point (displacement between the origins of T and C).

  • azimut (float | int | ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]) – The rotation around z of the camera. With an angle of 0, the camera would loot at the target point towards the positive x-axis of T.

  • elevation (float | int | ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]) – The rotation around the x-axis of the camera frame C. Note that if you want to lift the view angle, the elevation is negative.

  • fovy (float | int | ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]) – The field of view of the camera.

  • degrees (bool) – Whether the angles are in degrees or radians.

  • **kwargs – Additional camera parameters.

Return type:

MujocoCamera

Returns:

The custom camera.

jaxsim.mujoco.loaders.load_rod_model(model_description, is_urdf=None, model_name=None)[source]#

Loads a ROD model from a URDF/SDF file or a ROD model.

Parameters:

  • model_description (str | Path | Model) – The URDF/SDF file or ROD model to load.

  • is_urdf (bool | None) – Whether the model description is a URDF file.

  • model_name (str | None) – The name of the model to load from the resource.

Returns:

The loaded ROD model.

Return type:

rod.Model

Model#

class jaxsim.mujoco.model.MujocoModelHelper(model, data=None)[source]#

Helper class to create and interact with Mujoco models and data objects.

Parameters:

  • model (mj.MjModel)

  • data (mj.MjData | None)

Return the name of the base link.

Return type:

str

base_orientation(dcm=False)[source]#

Return the orientation of the base link.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

dcm (bool)

base_position()[source]#

Return the 3D position of the base link.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

body_names()[source]#

Returns the names of the bodies in the model.

Return type:

list[str]

body_orientation(body_name, dcm=False)[source]#

Returns the orientation of a body.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

  • body_name (str)

  • dcm (bool)

body_position(body_name)[source]#

Returns the position of a body.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

body_name (str)

static build_from_xml(mjcf_description, assets=None, heightmap=None, heightmap_name='terrain', heightmap_radius_xy=(1.0, 1.0))[source]#

Build a Mujoco model from an MJCF description.

Parameters:

  • mjcf_description (str | Path) – A string containing the XML description of the Mujoco model or a path to a file containing the XML description.

  • assets (dict[str, Any] | None) – An optional dictionary containing the assets of the model.

  • heightmap (Callable[[Union[float, Array, ndarray, bool, number, bool, int, complex], Union[float, Array, ndarray, bool, number, bool, int, complex]], Union[float, Array, ndarray, bool, number, bool, int, complex]] | None) – A function in two variables that returns the height of a terrain in the specified coordinate point.

  • heightmap_name (str) – The default name of the heightmap in the MJCF description to load the corresponding configuration.

  • heightmap_radius_xy (tuple[float, float]) – The extension of the heightmap in the x-y surface corresponding to the plane over which the grid of the sampled heightmap is generated.

Return type:

MujocoModelHelper

Returns:

A MujocoModelHelper object.

geometry_names()[source]#

Returns the names of the geometries in the model.

Return type:

list[str]

geometry_orientation(geometry_name, dcm=False)[source]#

Returns the orientation of a geometry.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

  • geometry_name (str)

  • dcm (bool)

geometry_position(geometry_name)[source]#

Returns the position of a geometry.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

geometry_name (str)

gravity()[source]#

Return the 3D gravity vector.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

is_fixed_base()[source]#

Return true if the model is fixed-base.

Return type:

bool

is_floating_base()[source]#

Return true if the model is floating-base.

Return type:

bool

joint_dofs(joint_name)[source]#

Returns the number of DoFs of a joint.

Return type:

int

Parameters:

joint_name (str)

joint_names()[source]#

Returns the names of the joints in the model.

Return type:

list[str]

joint_position(joint_name)[source]#

Returns the position of a joint.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

joint_name (str)

joint_positions(joint_names=None)[source]#

Returns the positions of the joints.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Parameters:

joint_names (list[str] | None)

number_of_bodies()[source]#

Returns the number of bodies in the model.

Return type:

int

number_of_dofs()[source]#

Returns the number of DoFs in the model.

Return type:

int

number_of_geometries()[source]#

Returns the number of geometries in the model.

Return type:

int

number_of_joints()[source]#

Returns the number of joints in the model.

Return type:

int

set_base_orientation(orientation, dcm=False)[source]#

Set the 3D position of the base link.

Return type:

None

Parameters:

  • orientation (ndarray[Any, dtype[_ScalarType_co]])

  • dcm (bool)

set_base_position(position)[source]#

Set the 3D position of the base link.

Return type:

None

Parameters:

position (ndarray[Any, dtype[_ScalarType_co]])

set_joint_position(joint_name, position)[source]#

Sets the position of a joint.

Return type:

None

Parameters:

  • joint_name (str)

  • position (ndarray[Any, dtype[_ScalarType_co]] | float)

set_joint_positions(joint_names, positions)[source]#

Set the positions of multiple joints.

Return type:

None

Parameters:

  • joint_names (list[str])

  • positions (ndarray[Any, dtype[_ScalarType_co]] | list[ndarray[Any, dtype[_ScalarType_co]]])

time()[source]#

Return the simulation time.

Return type:

float

timestep()[source]#

Return the simulation timestep.

Return type:

float

jaxsim.mujoco.model.generate_hfield(heightmap, samples_xy=(11, 11), radius_xy=(1.0, 1.0))[source]#

Generate an array with elevation points sampled from a heightmap function.

The map will have the following format: ` heightmap[0, 0] heightmap[0, 1] ... heightmap[0, size[1]-1] heightmap[1, 0] heightmap[1, 1] ... heightmap[1, size[1]-1] ... heightmap[size[0]-1, 0] heightmap[size[0]-1, 1] ... heightmap[size[0]-1, size[1]-1] `

Parameters:

  • heightmap (Callable[[Union[float, Array, ndarray, bool, number, bool, int, complex], Union[float, Array, ndarray, bool, number, bool, int, complex]], Union[float, Array, ndarray, bool, number, bool, int, complex]]) – A function that takes two arguments (x, y) and returns the height at that point.

  • samples_xy (tuple[int, int]) – A tuple of two integers representing the size of the grid.

  • radius_xy (tuple[float, float]) – A tuple of two floats representing extension of the heightmap in the x-y surface corresponding to the area over which the grid of the sampled heightmap is generated.

Return type:

ndarray[Any, dtype[TypeVar(_ScalarType_co, bound= generic, covariant=True)]]

Returns:

A flat array of the sampled terrain heightmap.

Visualizer#

Contents