GRID currently supports the following sensors with Isaac Sim.

​
Camera

This is a camera sensor that can be of type RGB or Depth.

​
Configuration

The configuration for the camera sensor can be set up in the the configuration file. The following parameters are available:

ParameterDescription
HeightThe height of the captured image
WidthThe width of the captured image
OffsetThe offset of the camera with respect to the robot’s frame of reference
Focal LengthThe focal length of the camera
Focal DistanceThe focal distance of the camera
Horizontal ApertureThe horizontal aperture of the camera
Clipping RangeThe clipping range of the camera

These parameters can be configured using the scene_cfg.yaml in the following manner:

sensors:
    - camera_0: # choose a name for the sensor
        type: CameraRGB # or CameraDepth
        config:
            prim_path: "" # Prim path to the link you want to attach the camera sensor
            height: 400
            width: 600
            offset: 
                pos: [0.0, 0.0, 0.0] # add init (x, y, z) position  
                rot: [0.0, 0.0, 0.0, 0.0] # add init (x, y, z, w) quaternion
            spawn:
                focal_length: 24.0
                focus_distance: 400.0
                horizontal_aperture: 20.955
                clipping_range: (0.1, 1.0e5)

​
Accessing Camera Data

The camera sensor can be accessed through the getImage method.

client = IsaacHumanoid()
image = client.getImage()

​
Contact Sensor

This is a simple contact sensor that works as a ray caster to obtain the proprioception values for all joints of the robot.

​
Configuration

The configuration for the height sensor can be set up in the the configuration file. The following parameters are available:

ParameterDescription
History lengthThe length of history to consider while computing sensor values

These parameters can be configured using the scene_cfg.yaml in the following manner:

sensors:
    - contact_forces: # choose a name for the sensor
         type: ContactSensor
         config:
            history_length: 3

​
Height Sensor

This is a simple height sensor that works as a ray caster to obtain the height values of points in a specified pattern.

​
Configuration

The configuration for the height sensor can be set up in the the configuration file. The following parameters are available:

ParameterDescription
OffsetThe offset of the height sensor with respect to robot’s frame of reference
Pattern ResolutionThe resolution of the grid pattern around the sensor
Pattern SizeThe start and end of the grid pattern

These parameters can be configured using the scene_cfg.yaml in the following manner:

sensors:
    - height_scanner: # choose a name for the sensor
        type: HeightSensor
        link: torso_link # link to which you want to attach the sensor
        config:
            offset: 
                pos: [0.0, 0.0, 0.0] # add init (x, y, z) position  
                rot: [0.0, 0.0, 0.0, 0.0] # add init (x, y, z, w) quaternion
            pattern_cfg:
                resolution: 0.1 
                size=[1.6, 1.0]

​
LiDAR

This is a LiDAR sensor that works as a ray caster to obtain a point cloud around the sensor.

​
Configuration

The configuration for the height sensor can be set up in the the configuration file. The following parameters are available:

ParameterDescription
OffsetThe offset of the height sensor with respect to the robot’s frame of reference
Pattern SizeThe start and end of the grid pattern
Max DistanceThe maximum sensing distance of the height sensor
Vertical FOVThe vertical field of view of the sensor
Horizontal FOVThe horizontal field of view of the sensor
Horizontal ResolutionThe horizontal resolution of the sensor
ChannelsThe number of channels used by the sensor

These parameters can be configured using the scene_cfg.yaml in the following manner:

sensors:
    - lidar: # choose a name for the sensor
        type: LiDAR
        link: torso_link # link to which you want to attach the sensor
        config:
            offset: 
                pos: [0.0, 0.0, 0.0] # add init (x, y, z) position  
                rot: [0.0, 0.0, 0.0, 0.0] # add init (x, y, z, w) quaternion
            pattern_cfg:
                channels: 16 
                vertical_fov_range=(-15.0, 15.0)
                horizontal_fov_range=(-180.0, 180.0)
                horizontal_res=0.2
            max_distance: 100

Was this page helpful?

Previous
Training Workflow
Next