Torque and acceleration are proportional to current, while no load velocity or RPM is proportional to the voltage which may reduce some 20% with rated load in a good match.
Analogy for Speed control
Consider the source impedance of a motor and mechanical load impedance toa voltage of a power supply. Since torque is current adding load reduces the RPM is the ratio of motor torque which is reduced to high impedance due to back EMF cancelling the applied voltage. So it is like a power supply with the variable impedance that rises when the load reduces. So the speed regulation can be computed from the current sensing and the commutated current is the RPM so speed is automatically available without a Hall sensor.
ESC vs PID ?
The best control system is a 1st order where the feedback is the same as what you control, rather than the integral then requiring a derivative to anticipate and thus you have a PID control system with compromises for speed and overshoot.
To be or not to be (controlled)
This means current control is the best for phase margin, overshoot and stability, but usually, you want to control speed, not acceleration. So that means Voltage control/Overshoot is minimized by avoiding a current step and using controlled ramped currents and a controlled acceleration and braking ramp or simply an active current limiter is needed.
Motor start current is V/R the DC resistance of the motor coil or DCR as they are all specified. Imax=V+/DCR, which can be up to 10x max rated load current!!. Thus ramped-current control can limit this overcurrent using a Vpwm/for Vavg/RPM system control.
The good thing is sensing current can be used to commutate the windings by detecting the position of the magnet's effect on current and thus adapt instantly and possibly phase lock to this detection.
The controller always knows what sequence to use for 2phase or 3 phase etc. BY DESIGN.
Hall alignment or dead fans?
The issues could be, Hall sensors add cost and shift slightly with temperature. When they are too close to the magnetic polarity switch they go back and forth in "some" fans or stall until pushed (once in a while) up to 1% from my previous bad suppliers (even NIDEC). Perhaps temp rise shifts the magnetic strength of one magnet strength more than another which shifts the midpoint. ( magnet tolerances)
In motors, this method allows no mechanical issues such as Hall sensor position alignment with the magnetic reversal. There may be mismatched winding inductance and resistance issues that must be considered but it can definitely start and stop easily.
The other problem is high power ESC motors if commutated too early apply torque in the wrong direction for that 0.x % of the time if switched too early. So a deatime is used to coast thru that ambuious transition, then you waste some magnetic torque. Its a very tight tradeoff and critical tolerances are needed.
The question is how efficiently it commutates close to the magnetic pole reversal, what happens with load disturbances to current and commutation and is it scalable to large machines.
This is part of a larger modern control method called Vector Motor Control which you can research.
I clearly state this is NOT a finished product, but a tool to gain insight into the finer details on this method of control for different dynamic loads and motors.
Read the fine manuals.