Optical systems like Vicon, Motion Analysis or Optitrack have obvious calibration methods but when it comes to calibrating inertial mocap systems, good luck. You will need lots of tools, and then some you need to be good at since they are not 100% automated.
Synertial has taken a good stab at it though.
Let’s start with explaining our Calibration Jig and go from there.
Our latest generation Jig is a simple plastic device which allows hands to calibrate in the same position as previous models like the calibration mitten but in a much much smaller form factor
We now need finger lengths as well as the posture (the jig forces the X and Y axes and the lengths are in Z). We still need Z data for the skeleton.
We use our office copy machine to scan my hand in an A4 size scan and save in .png format. I load that scan into Synertial’s Computer Vision SW, Kinexact-Hand, to extract some lengths for the skeleton segments’ Z axis.
The red circles are for drag&drop editing if needed but all I do is click ‘Save file’.
Then I go to the SynDash Skeleton Editor SW and load the skeleton file that matches that particular Mitten design (so I get X & Y), and then use Kinexact-Hand to get the Z data.
Once the 3D model has its XYZ and the Jig's worn, they can be linked by a keystroke.
One more thing to calibrate…
There is another element that needs to get calibrated, simultaneously, which is the establishment of the forward direction. Synertial does this by designating one of the glove sensors as the ‘Reference IMU’, which the system will look at during an ‘arcing’ maneuver around ‘X’ axis (to establish a Sagittal Plane), to see which way is front.
The arc is made between 2 postures we refer to as POSE-1 (or sometimes call ‘X Reference’) and POSE-2 (also called Zero pose).
Some basic ways to calibrate in 2 poses: