I think you should go simple on this, first, closed loopback system need stability and there nothing stable in a motor going from stop to high rev with non-linear torque force on it, that why even in slower system like an engine they use mapping that run open loop in fast changing condition.
Second, I think, if you are afraid of restriction, you shoud use a bypass system, it does not take away any energy from the engine when you will not be using your supercharger.
Third, be careful about technical spec on component, MOSfet are not good for most motor control appication Ton-Toff are normally to long(so they heat up to much), use IGBT its like MOSfet but design for this kind of application.
I would go with a simple 8 bit PIC familly microcontroler(microchip), they simple to code, programmer dont cost much, and are stable in a wide range of temp(summer heat in car can get pretty high), oh and do not use a crystal for the clock use an oscillator, vibration will kill the crystal(made this error once and you do not want for full uncontrolable power on the elec. motor).

I least that a few thing I would think about if I would go into that kind of project.

This is the one point I would argue about. Fenton has found that having the system set to run at low level when not engaged does not drain the batteries and by the nature of the beast puts a very slight amount of pressurization on the system that makes a noticeable improvement in throttle response. Plus we are trying to keep the install as affordable and simple as possible. We want to avoid adding any more hardware than is absolutely necessary.
If we can figure out an affordable way to use existing electronics (Unichip?) rather than reinventing the wheel, that would make sense too.

It is an automotive engineering exercise that depends greatly on programming.

Tell me about it. The last time I earned a living programming, it was in cobol.
I play with current stuff just so I'm not completely out of the loop.

I think people are sticking with what the know and we have a lot of programmers. I'm focused on hardware and have parts on the way to start building a prototype. After all programming does nothing without something to test it on.

I'm in total agreement. The closer this is to a stand alone system to start with the better. I like the idea of a bypass, but I'm getting hung up on what to use for the valve.

Two bypass options as I see it:

1. An electrically controlled valve. Using a throttle body is what first comes to mind.

I don't personally like this. It becomes another active component that adds to complexity.

2. A spring loaded valve that is is held open when there is lower pressure on the engine side relative to the intake side. As soon as the compressor kicks in and there is higher pressure on the intake side the valve is forced closed.

Some sequential bi-turbos use this type of valve but I'll be darned if I can find where I can buy one.

An ebay BOV with a very weak spring would work, no?

under, say 30in/hg - 5in/hg it can be open. 5in/hg to boost it would be closed, and there would be overlap (programatically) with the compressor kicking in?

The boost pressure along with the little spring pressure should keep the diaphragm closed under boost....I think.

I've had two turbocharged cars and never held a turbo in my hands so I'm having trouble picturing the exhaust housing and impeller.
If you drove the exhaust side with an electric motor, would it compress air? If it would, couldn't you just use one of those with a built in wastegate?

Edit: never mind, even if it worked, it would be horribly inefficient. The housing is clearly designed to encourage airflow in the other direction.

Its a mixed project. There are four main parts to figure out

1) The supercharger hardware side (Compressor/Speed controller/Intercooler?)
2) The fabrication side (Mounts, pulleys, brackets, braces)
3) The electronics side (Capacitors, charging system, discharge system)
4) The controls side (Software, sensors)

I don't think the exhaust wheel is designed to compress air, more of a water-wheel type of deal. I could be wrong.

I not against the idea, just saying if he afraid of some restriction that would cause a lot of loss power, this would be more efficient way to do it. When i say efficient, I mean, longevity and mpg and electrical charge on the alternator/regulator and so on, I do not know want kind of motor he will use, but spinning continuously a high rpm motor could possibly not be efficient in term of live.

It would be turning at a small percentage of max most of the time.
To be honest, though, this thing could turn out to be affordable enough and easy enough to work on that if I knew I would have to replace the motor every two years, it wouldn't upset me.
...and I really like the idea of improved throttle response.

I looked up some images. The wheel itself had a reasonable shape, but the housing does not look like it would work well going in reverse flow direction, which is what it would require.
Oh well, sometimes when you think outside the box, it falls on your head.

I would start by just doing the supercharger side of the project, its already a big thing to have that design and working properly, the rest can be added later when proof of concept is achived.

True enough, we are kind of putting the cart ahead of the horse.