The position read by the encoder is relative.

[jakechung42]

The reading of the encoder is relative. By saying this, I mean that the encoder stores the pulses in its object to count the angle. If the object is reset by means of resetting the code or resetting the power, it will begin to count again from 0. This causes the actual angle of the top plate to be not the same as the reading of the encoder. An idea is to mount an IMU to the platform and create a routine to calibrate the platform.

[jakechung42]

Using an IMU wouldn't work either because IMU position is also relative. I think the best way to do it is to use a potentiometer for calibration but using an encoder for sensory measurement in the feedback loop. Initially, the pot wasn't desirable because the voltage range is only 5V so the pot resolution wasn't really good and I didn't want to change the operating voltage, so I went with the encoder. Using the pot for the calibration shouldn't introduce the resolution problem again.