Just because it is not directly measuring where the sled is does not mean that it is open loop control. What matters is where the laser is in relation to the tracks on the disc, where the sled is doesn't matter - unless it's run into the end of travel, and there's a switch for that - so that isn't the variable you close the control loop around.
The tracking needs to be done to less than 1μm so is not possible without feedback from the optical system. The optical signal is split into side beams and these are compared, this then can control both the micro system adjustments and the larger sled motor movements.
The side beams created by the diffraction grating are positioned forward and back of the main beam straddling the track of pits being followed (not directly on either side as shown in the diagram - but that was easier to draw!).
Segments on either side of the photodiode array designated E and F monitor the side beams. Tracking is perfect when the E and F signals are equal.
So if you are oscillating a little bit off the track, then you can use the tracking coil to compensate, but if you trend continuously to one side you would move the sled. I don't know how this is done, but that looks like a job for a Kalman filter. The linked page also mentions open-loop heuristics for track seeks, but then you check the timecode where you landed and go back into closed-loop fine control.
The principle here of 'measuring the effect rather than the actuator position' is commonly used in both hobby and industrial settings; it arguably is what makes closed loop control closed.