🌡️ Sensor#

By configuring sensors, users can more conveniently obtain the state information of physical objects, such as position, rotation, velocity, acceleration, etc. Sensors do not affect the results of the physical simulation and can be attached to different objects, such as bodies, sites, joints, etc.

📋 Sensor Overview Table#

Sensor Type	Measured Object	Coordinate System	Return Dimensions	Main Applications
accelerometer	site linear acceleration	site local	3	motion perception, fall detection
velocimeter	site linear velocity	site local	3	velocity monitoring, motion control
gyro	site angular velocity	site local	3	attitude control, motion estimation
jointpos	joint position/angle	joint axis	1	position control, limit monitoring
jointvel	joint velocity	joint axis	1	velocity control, motion monitoring
framepos	object position	specified reference	3	relative positioning, path tracking
framequat	object orientation	specified reference	4	attitude control, orientation monitoring
framexaxis/yaxis/zaxis	object axis	specified reference	3	direction analysis, alignment control
framelinvel	object linear velocity	specified reference	3	velocity tracking, motion analysis
frameangvel	object angular velocity	specified reference	3	angular velocity control, rotation monitoring
framelinacc	object linear acceleration	specified reference	3	acceleration monitoring, dynamic analysis
subtreecom	subtree center of mass	global	3	stability analysis, COM tracking
subtreelinvel	subtree COM velocity	global	3	momentum analysis, motion monitoring
subtreeangmom	subtree angular momentum	global	3	momentum conservation, rotation analysis
touch	site contact normal force	site local	1	tactile perception, collision detection, force feedback
contact	geometric contact info	local contact frame	variable	ground contact detection, grasping force feedback, collision analysis

🚀 Quick Start#

Sensor Configuration Example#

Sensor configuration examples can be found in examples/assets/site_and_sensor.xml:

<sensor>
    <!-- Local coordinate system sensors -->
    <accelerometer name="acc_0" site="imu_0"/>
    <velocimeter name="vel_0" site="imu_0"/>
    <gyro name="gyro_0" site="imu_0"/>

    <!-- Joint sensors -->
    <jointpos name="joint_pos" joint="shoulder"/>
    <jointvel name="joint_vel" joint="shoulder"/>

    <!-- Reference frame sensors -->
    <framepos name="global_pos" objtype="body" objname="robot" reftype="world"/>

    <!-- Subtree sensors -->
    <subtreecom name="robot_com" body="torso"/>

    <!-- Contact sensors -->
    <touch name="palm_touch" site="palm_site"/>

    <!-- Contact sensor (detailed) -->
    <contact name="box_floor_contact" geom1="bar" geom2="freebox"
             data="found force pos normal tangent" num="4" reduce="none"/>
</sensor>

Note

MotrixSim Limitation: Currently does not support cutoff, noise, and user attributes from the MJCF standard

Python API Usage#

Get data from specified sensors:

# Get sensor value directly
v0 = (model.get_sensor_value("vel_0", data),)
v1 = (model.get_sensor_value("vel_1", data),)
v2 = (model.get_sensor_value("vel_2", data),)
v3 = (model.get_sensor_value("vel_3", data),)
(print(f"velocimeter values are : {v0}, {v1}, {v2}, {v3}"),)

acc_0 = (model.get_sensor_value("acc_0", data),)
acc_1 = (model.get_sensor_value("acc_1", data),)
acc_2 = (model.get_sensor_value("acc_2", data),)
acc_3 = (model.get_sensor_value("acc_3", data),)
(print(f"accelerometer values are : {acc_0}, {acc_1}, {acc_2}, {acc_3}"),)

gyro_0 = (model.get_sensor_value("gyro_0", data),)
gyro_1 = (model.get_sensor_value("gyro_1", data),)
gyro_2 = (model.get_sensor_value("gyro_2", data),)
gyro_3 = (model.get_sensor_value("gyro_3", data),)
(print(f"gyro values are : {gyro_0}, {gyro_1}, {gyro_2}, {gyro_3}"),)

Detailed Sensor Documentation