Wiki ยป

[0: Preparations]

The steps presented here require an assembled motion capture helmet, and a working robotreality software installation.
If not done yet, please complete the previous steps:
  1. Download, Building and Installation and
  2. Setup & Calibration.

If the software was built with Flobi support it allows direct emotion transfer to a real or virtual Flobi robot. Please note that this software part is not yet freely available. Without the Flobi support the software will only write the data to a CSV File. If you want to adapt it to your own robot please have a look at DataOutput*.cpp.

[1: Face Markers]

The facial expression tracker expects colored markers on special positions in the users face. The best colorchoice is fluorescent green. We use neon green post-it note paper glued on skin friendly adhesive paper as found on ECG stickers.
Please place the markers on the positions shown in the following picture:

Some hints:
  • place the two nose markers on places that do not move during normal facial expressions
  • eyelid markers as close to the edge as possible (and comfortable)
  • mouth markers should be arranged pairwaise on top of each other vertically

[2: Robot setup]

This step only applies when working with the real robot. If you compiled without Flobi support (default) the data will be written to CSV file instead.
Connect the Flobi robot to the Host PC and run the XSC2 Server with the correct configuration file for the robot. Switch on the motor power and initialize and calibrate all joints as usual.

[3: Motion Capture & Live Playback]

Start the robotreality software with the appropriate config file:

/tmp/robotreality/mocap/$ robotreality_mocap config/live_ttyUSB0.mcfg

All config options can now be set up as described in Setup & Calibration.
The only difference is that you tell the user to look straight when clicking the head zero button as described in section 5.

Once you check "[x] data output enabled" the data is either sent to a CSV file or, if configured, to the real robot. Please uncheck "[ ] show results" in order to reduce latencies and to reach the full 60 frames per second updaterate.

[4: Known Problems & Solutions]

If the software fails to detect the XIMU sensor please first check that it is connected to the right ttyUSBx. At least under Ubuntu precise there is a known bug where the Modemmanager grabs all ttyUSBx devices and fails to release them properly. See the libXIMU notes section for details.