New motor drive method for Instaspin program

The improvement shown with this method is in comparison to a non-PWM 120 degree (Trapezoidal 6-step commutation). It looks to be an improvement but you could also look at doing true FOC. That would be my first inclination. We have had excellent success using IS-FOC on high speed motors to drive best dynamic response and excellent efficiency.

One of the challenges of course in this sort of application (low voltage / high current) is in dealing with the current measurements. For FOC you require 2 or 3 (single shunt reconstruction is possible but much more difficult and often not worth the effort). Because of the high current the shunt approach can become expensive (and result in efficiency losses) almost to the point that phase current measurements using hall sensors (LEM is a popular brand) start to make sense. But this is at cost increase to the bus current measurement approach that is standard in 6-step methods. Also, with FOC, to get maximum bus utilization to compare to full BLDC output power you need to be able to over-modulate. To still take current samples for the feedback in this area you will require the 3rd shunt (or 2 phase). Again, more cost.

Back to your original question: Is it possible to have this "Control of current conduction angles and phases" without position sensor?

I believe of course the answer here is YES. Considering that the profile for this motor is of normal operation at higher frequencies, a Bemf (phase voltage) based observer could be used. Something like InstaSPIN-BLDC would be very robust. This would give you the commutation states. However, in the paper's implementation the modulation/commutation is adjusted so that for 2/3 of the duty cycle you aren't actually PWM'ing the FET, but just leaving it full on/off.