Hey guys,
I'm working on a little DIY project in which I intend to make my own dataglove. Now I'm kinda stuck on choosing the right microcontroller. And since it can be quite costly, I thought to ask for some help.
So: I'm going to use 13 flex sensors (maybe 14) and 1 IMU (probably the 6050). now the problem is that I'll need a total of at least 14 AI.. And the small controllers only have about 6.
The controller can't be too big for dexterity reasons and also will need the availability to add a Bluetooth module.
I've been looking at some multiplexing/shield options but I'm just not confident (and rich) enough to just start ordering and see where that leads me. Also the multiplexing is not (yet?) which makes the decision making even harder.
So if any of you guy's got an idea on which configuration can be feasible, feel free to let me know! 
Peace out,
Lars
Which exact sensors were you planning on using? My guess is that 13 of those are going to be the majority of your cost! 
MCUs are relatively cheap, and some of the Teensy boards have >13 analog inputs.
I built a glove about seven years ago, and decided that the flex sensors were an extravagance I couldn't afford. I bet they're more expensive now than in the 1980's (adjusting for inflation), and they don't seem to have advanced at all since they were used in the Nintendo Power Glove. At the time I went looking, all of the flex sensors I could find were like 3" long, too, so not really appropriate for per-knuckle use. I ended up using short lengths of stretch resistor cord instead. That stuff was (relatively) inexpensive but had interesting hysteresis issues that required more filtering than would have been ideal for the framerate I was hoping to achieve. Still, coupled to an ugly OpenGL model, it was a lot of fun to play with.
It'll be the flex sensors from Flexpoint. They are way more precise compared to other manufacturers like Spectra and come in standard sizes of 1", 2",3" and 4". And the 2" ones that I'll use are just $7. It's still a cost so I try to cut the cost wherever I can.. You should check them out!
Yeah, those are the ones I was talking about--the Power Glove sensors. Good to see that the smaller sizes are more available now! But at $7 each, that's $100 worth of sensors, and an MCU module is going to be somewhere in the $20's or $30's, probably. So again, the sensors are your big cost (ignoring time, which is only ok to do if this is a for-fun thing :)).
Two inches is still pretty long if you're planning on doing this per-knuckle. I suppose I'd be more tempted by the one-inchers, but then I don't know how you'd planned to mount them.
The first thing I tried, which was a humorous failure, was to make my own paint-on stretch sensors using a mixture of carbon black and liquid latex. The idea was to run wires to either side of each knuckle and then paint over them with this stretch-resistor rubber mixture. The concept was proven, but the resistances were so high--100's of k to megs--that I back-burnered it. Might have to try that again later with a different rubber. Eventually, in an effort to get the resistance down to something more usable, I over-saturated the latex and it just started crumbling.
Here's an (untested) idea for a really cheap DIY capacitive sensor:
Let two tin foils overlap, isolated by some non conductive foil, so that they form a variable capacitor. Mount the conductive foils to both sides of a knuckle, so that they overlap more or less when you bend the finger. Then measure the capacity difference, using the CapSense library or, better for many such sensors, with some specialized chip.
The capacitance change due to the distance between the foils will be much stronger (and non-linear) than the capacitance change due to overlap area. Any tiny inaccuracies or wrinkles will screw the results.