Yes, repeated calibrations may result in slightlydifferent baseline levels every time you run the program. These changes would be due to changes in environmental conditions, electrical noise and sampling noise. However, it is unlikely that this will change significantly the behaviour of the touch tracking across participants: the differences in finger size and shape and in the way they placw the finger on the sensor across participants are likely to affect the readings much more than the baseline calibration value.
Anyhow, you can access the current calibration level by setting the device into BASELINE mode and then performing a read of the raw values. If you see significant differences across calibration cycles and you really want to use the same calibration for all your participants, you can set the device into RAW mode, read the values after setting all other settings and use those as baseline (you should keep these across runs). Then at runtime keep reading in RAW mode, take the difference between the readings and the baseline you cached and feed the difference to the CentroidDetection class.
