When I read your long story of building this hovering, autonomous, 3dcPos tracking robot, I had to log in.
Although I'm fairly new to micro robotics, I have some background on aerodynamics, and anything related to flying objects, and experienced in R/C aircrafts.
I'll be honest and mean. It's not gonna work, and you'll fail miserably. Period.
People have been travelling with wheels for thousands of years, and it took another 300years after the first controllable vehicles came out that the wright Bros manage to stay in the air for 13seconds.
It's only been a century since the Flyer-1 took off, and it's only been last decade or so that we started seeing robotics even DARE to mix with aircrafts (by institutes who specialize in these things).
From my experience of watching what others(specialized in at least University level of aero studies) are capable of, it'll take another 5-6years before we start seeing some form of autonomous robots hovering over our heads.
From what I hear (your CAD doesn't display for me), your robot is considered a micro~R/C-aircraft. That is, anything small enough to fit in your bag. Here comes my batch of reasonings as to why you should consider things beforer you even start this project:
Firstly, one of the major reasons why robotics couldn't mix with micro-aircrafts is due to source of power. Gravity is a mean thing. Just to keep your robot off ground, it takes tremendous amount of energy. Those R/C crafts you see have Battery life of at most 20minutes. This is done by specially designed batteries made specially for toy companies. (wowee uses a battery size of a bubble gum to power their dragonfly for about 8minutes). Even from annual R/C convs, most battery powered planes can't stay up more than 15minutes (that is, if the wind conditions are optimal), and that's why we have fuel combst engines. Considering that you want full electronic components and use direct thrust to gain lift AND hover, you need some NASA level of engineering to do all these things that you want it to do.
Secondly, using twin rotors sounds fancy, but to have all your 3 dimensional controls with pure thrust will require some 5mb big coding to sense and calculate your moves. This isn't like a wheeled bot where you're always on a stable leveled horizon and you move forward and backwards. You get constant fluctuations and dynamics in all your coordinates. ALL outdoor aircrafts require either extremely stable design, or sensitive sensing and sophisticated programs to keep their head up and stay balanced. Your controls will not work as simply as you think it would.
Right off back of my head, I can think of dozens of reasons as to why this will not work(GPS?/mapping radar for in a 3D situation?/Autonomous only works in static env where data programmed is all it needs, not in a always changing environment/mass(material)/where are u even tryin to get thrust for all this?/money???/how many hours will you spend...?/Many other "roflable" ideas. , but I don't think it's even worth it to fight over such minor post...
It seems like your dedicated, but sometimes you have to take your time and practice before your can jump over a wall.
However, I offer these tips in case you are really determined.
1) To Hovering problem, you can use whatever you want, just as long as your thrust is greator than your weight. This problem is rather easy... One propellar or twenty does not matter. Just like the vid of airplane hovering in mid air, if your bot's weight is heavier than radial drag your propellars create, you won't spin (good example would be drilling a hole on a wood-your body mass/innertia VS drag is too big that system will try to do least work by not "rotating" you). You can even have jet-engine for all I care. Theoretically, its the motor's power that determines take off weight, not the RPM or the Torque.
If you increase your torque, you reduce torque, and more speed, less strength. Yes, there is the optimal line because of the mass of the propellar and the minimal rad drag, but you can go with 25000 RPM small motor+3cm radial prop and achieve same thrust as to much powerful motor with less speed and more torque with bigger props (bigger doesn't mean better).
2)As to controlling in z(height) axis, I strongly suggest using speed of rotors. Pitch control requires high level of engineering. To have pitch control in such small robot packed with electronics is going to be extremely hard. If you want to use pitch control, you need lines of calculations as to shape of your prop and dynamics involved.
3)Turing- You said one can turn slower and one faster, but the problem involved with coaxial rotors, is that once your reduce speed on other rotor, you get difference in lift. This causes fluctuation in altitude. Helicoptors/VTOL like chinook and osprey use sophisticated level of programming to deal with this variance. if you are building a helicoptor, I suggest you get control for height and X/Y controls seperately. This means get vertial climb control seperately and turns/forward backwards seperately. However, if you MUST have it your way, use dividing thrust vectoring component under your "lift" rotor. In case of osprey(VTOL craft picture), it uses it's Ailerons to convert downwards thrusts' to sideways. You can use vector thrust like F-22 Raptor or ailerons, straws, or anything else that converts downwards thrust to sideways.
4)As to foward, I repeat Vertical control fluctuates and ur X/Y (forward and sideways) are not. Rule of the book is to have Vertical and Forward/turn seperated. You will find that Vertical thrust and maintaining that value will be much tougher than X/Y, as gravity only affects downwards. And if you get a sideways flux (like wind blowing from sideways), your bot will tild, creating diversion of your vertical thrust into sideways and some crazy movement (most cases, strong winds for R/C helicoptors result in "Impossible to control scenerio", where as in airplanes more stable). But if you really insist, then you can use diviersion of vec comp of lift thrust, as 3). Using some kind of rudder/elevator/aileron to shift winds coming downwards is easiest way if you choose to stick with your design.
Honestly, I feel like laughing at you, as to a fool who haven't built a 50$ bot to tackle such challenging project.
Seeing that you find the vid of acrobatic R/C plane amusing, you have a long way to go. Start small and get a R/C plane. Understand the physics involved. Starting at a NASA or University research level project is like skipping over the basics of skating, and going to the NHL, just cause you love hockey.
Hope I don't discourage you too much. Someday you'll be able to build your bot. Just that, that someday is not today...