WOW, that video looks great. I've always wanted to program/design my own AC/brushless servo drive but don't have the programming knowledge. Are you planning on releasing the code open source? That way more people could contribute and develop extra features for the drive.
There are no limits to how sophisticated a servo drive can be. All the industrial versions now feature auto-tuning and sophicated control beyond PID. While I don't expect any of that on a simple hobbiest level drive, it does make a great PH.D thesis.
Have you thought about driving the motors without the hall sensors? I'm not sure how your present start algorithm works. But it would be really cool to use "wake and shake" to initilize the encoder to the rotor position. This method is very appealing because it makes an AC drive only 1 wire more complicated than a DC brushed drive. The only downside it the short 2 or 3 seconds to wait for the controller/motor to get synced up. Kollmorgen also has a method that slowly energizes one set of coils. When the rotor starts turning, it moves the commutation point to prevent further movement. This occurs until the rotor and field are aligned. This way the rotor only moves a few encoder counts on start up. Does any of that sound "doable"??