CloPeMa Sensors

This document lists all the sensors present on CloPeMa robots, their output and their possible application.

RGB and depth (Kinect like sensors)

Kinect like sensors uses a principle of structured light to capture depth. A known pattern is projected on a scene using IR diode and captured using IR camera, the disparity between known and captured pattern are used to compute the depth. Along with IR projector and camera these sensors are equipped with RGB camera.

Variants

- Xtion PRO Live (Xtion)
- PrimeSense Carmine 1.09 (PS)

Output:

- RGB (640x480 pixels, 30 fps)
- IR (640x480 pixels, 30 fps)
- Depth (640x480 pixels, 30 fps)

Application:

- Fast scene recognition
- Grasp point detection
- Colour and depth segmentation

Stereo head

Output:

- RGB Left and Right (4928x3264 pixels)
- Depth

Application:

- Scene understanding
- Colour and depth segmentation

Photometric Stereo

Output:

- 4x BW
- Albedo
- XYZ Normals
- Depth
- Point Cloud

Application:

- Fabric type recognition

Force Torque sensor (Only CVUT)

Output:

- XYZ Force [N]
- XYZ Torque [Nm]

Application:

- Detect contact between a gripper and a table
- Stretching of a garment
- Measure the weight of a garment

Tactile sensor on gripper finger

Output:

- 16x pressure

Application:

- Fabric type recognition
- Seam following

Light sensor on gripper finger

Output:

- Single value

Application:

- Check of the garment presence
- Some clue for the fabric type recognition

Microphone on gripper finger

Output:

- Sound

Application:

- Fabric type recognition (with rubbing)

Rotation encoder on gripper jaws

Output:

- Angle of the gripper jaws

Application:

- Check contact of the gripper with the table
- Stretching of a garment