My wife can no longer type on a keyboard, and uses a mouse to type with an on-screen keyboard. I designed a set of buttons for her, so she uses mouse movement in one hand, and buttons in the other, but I really want to provide her a better and less-grueling option. (Consequently, she breaths on a ventilator, and no voice-recognition software has ever worked for her, aside from the impracticality of doing ALL typing, with others around, by voice.)
Now, I've been doing machine-learning/neural networking for over a year, and I'm idealistically thinking that I might be able to place a bunch of capacitive proximity sensors in a palm-held device, perhaps 10-16 of them, and just take this data into the computer for a [potentially elaborate] processing. I have a bit of experience with electronics; a couple years ago I tried making a plain button set with capacitive sensors inside:
Without active shielding, I got way too much interference between the sensors (and likely their wires). The wires ran to an AVR board (a promicro16), ie. not a sensor controller and, as you can see, I couldn't come up with any useful way to route them (as nothing would solve the interference problem without proper shielding). With the AVR, I didn't have a good way to drive active shields.
So... I'm looking to TI's products for a solution, and you guys/girls for help. :)
1. I need help selecting a suitable controller(s) for the proximity sensing with active/driven shielding.
2. Sensor options:
2.a. I need to find very inexpensive flat sensors (or flexible) that won't interfere with each other (and therefore supporting active shielding?) These might go on top of her current mouse, under a hand-support I made, so their thinness is very important. Here's a photo of the hand-support made for her mouse (it lets her use that mouse in her right hand, for movement, without accidental clicking of the buttons, which she doesn't do with her right hand anymore). www.flickr.com/.../
2.b. I can just custom make sensors with stacks of copper tape+insulators. For the leads, I was thinking headphone wire might suitable, driving the stranded ground as active driven shield?
3. I need to get the data to the computer. With the promicro16, I had a usb connector, and used the LUFA library, but I didn't get far -- only doing terminal communications when developing. This worked in linux but was cumbersome in Win7 because each time I connected I'd have to re-set the COM parameters from cmd shell (windows wouldn't remember the settings no matter what I tried).
4. Any other help! :)
Thanks.
ps. She's using a button set with microswitches in it currently. The tactile feedback was helpful for her. The purpose of the capacitive sensing was originally to give her a simulated mousewheel (calibrating to her thumb position, then slight pressure or release would scroll up/down). In any case, even that was put on hold. Nevertheless, for the more advanced sensing of hand/finger motions for typing, I am absolutely going to require something more elaborate than switches. Here is the design of the button-set she's actually currently using: www.flickr.com/.../
The wires just run to a USB mouse, so nothing I'm currently doing is using my own USB device.