Well, it depends on what you’re trying to accomplish.
For me, Something like the laser interference described here would work effectively because I have a really large area and a really shallow depth, but for my purposes, I’m not looking for a multi-touch solution, just a natural feeling interface, and I’m looking to have clean lines and keep the sensors away from the actual touching.
In short though, embedded systems refer to discrete components designed to perform a single task (usually under some kind of time constraint) within a larger item as a whole. I wrote up a quick overview in the Beginners area that roughly describes my plan. http://nuigroup.com/forums/viewthread/23093/
You need to look at your objectives and find out what you’re building, its purpose, and break it down from there. If you’re thinking of using capacitive multi-touch sensing, you’re going to have a lot of stuff on your hands, because it would be complicated. You’d need to hook up an oscilloscope or logic analyzer to a working screen, and monitor the signals being sent from the screen to the controller, including the initialization codes. Then, you’d need to correlate those signals to actual inputs. Finally, you’d need to develop a driver for your board which would work with these signals.
Like I mentioned, I’m planning on using the PIC16F707 microprocessor as each has the ability to handle up to 32 channels of a single capacitive change, and only 2 at one time due to the limits of the IC.
What this means though is that I have static displays, that will never change once its in. On the back of each ‘button’ will be a metallic pad with a small air gap, so that touching the plexi in front of the pads will cause a capacitive change that the PIC’s will register.
Using a Raspberry Pi as the controller because the extra computing power needed for the image recognition is no longer required, the Pi will aggregate signals, parse them and provide haptic feedback through sounds or whatnot.
If it sounds like a lot of work - it is. It’s also a far cry from being efficient. If I was designing this for a client, I’d use an LCD panel and a single custom touch screen, but you’re moving into the thousands of dollars/euros for the screens, and then the programming behind it all.