Damien Lebigre
Email: MailTo(damien.lebigre AT gmail DOT com)
AttachInfo AttachList [attachment:1stAttempt.pdf] [attachment:BicubicInterpolation.pdf]
In this page, you will be inform about the project I am involve in. You have, thanks to this page, the opportunity of following step by step this amazing project called Belexes and sometimes download some files that will illustrate my intention.
This project is, actually, for constructing a multipoint, three dimensional, hand-controlled interface system. I am working with Bill Evans.
In order to know me well.
I am, actually, a student in an engineer school in France called [http://www.ifips.u-psud.fr/fr/specialites/optronique.html IFIPS] . I have the opportunity of making an internship for 3 months ( I have started to work in the Center for Music Technology on the 5th May 2008). I have choosen to make my internship in this department for several reasons:
- I am totally fond of Music and High Technology and this internship offers me the opportunity to mix both.
- I am really interesting in all the touch interface technology and thanks to this internship I can increase my knowledge about touching interfaces.
- Working in an english spoken country in order to increase my adaptability and my english knowledge.
Now go back to the project.
Theory Around This TouchScreen
Before the beginning of my internship, I have read several scientific papers about [http://www.billbuxton.com/multitouchOverview.html Touchscreen Technology]. By reading all this paper, I have noticed that nobody has yet created a commercial product using our technology. It seems that only SONY has created a prototype using the same technology that we use: [http://ftp.csl.sony.co.jp/person/rekimoto/smartskin/ The SmartSkin]
The technology involved in the project is the capacitance effect. Thanks to that effect, we will be able to detect how far is the hand from the table.
How the TouchScreen Works ??
Before going further in the project, I have to explain how the project works. To measure how far away the hand is, I have already told you that we are using the capacitance effect. In order to involve it, we have to build a grid. The vertical wires are the transmitter wires (the ones which have the signal). The horizontal wires are the receiver wires (the ones which get a singal thanks to the capacitance effect). The receiver wires are actually the ones which will measure how far is the hand from the table.( ![/!\](/moin_static/modern/img/alert.png "/!\"){kind=small}
in order to have a capacitance effect, there is absolutely no contact between the receiver wires and the transmitter wires ) The measurement points are the intersection between the receiver and the transmitter wires thus for a 8 per 8 matrix of wires, we have got 64 measurement points. After cleaning , getting and converting the data from each point, an image processing will determine more precisely where the finger is.
Dairy of My Work
3 first Weeks
First, I have read a lot of papers so as to know how to build this project. It was a big part of the work at the beginning. Given that we are creating something that have nearly never been done before, we have to spend hours searching for the best way to do the first prototype. Then we have tried to test the capacitance with a small experience (see 1st attached file)) and it works well because we could see a decrease in the signal.
In order to find where the finger is we just have to make a picture processing. For that I have written a small code in Matlab (see 2nd Attach file) (which will be probably converted into a C++ code in order to be faster). This picture processing use a [http://en.wikipedia.org/wiki/Bicubic_interpolation Bicubic Interpolation Proceed]. You could find the code in the attached file. For the test code, I have created a square matrix (which is actually a .xls file). Each number in the matrix is a digital value, the higher this value is, the further away the finger is. We can thus use .xls files in order to save the data. This code could increase the sensibility of the whole touchscreen. After this little work on the image processing, we thought about the way to make the data processing and how to create the reference signal.
For creating the signal and making the measurement, we have chosen to use micro-controller, because it is very adaptable to our needs, we can be programmed pretty easily and it is cheap. In order to program the chip, we will use [http://winavr.sourceforge.net/ WinAVR] which is a software that can translate a C-code into Assembler. It is an open-source software. This is not the one I am used to use so it will be very interesting for me to use this.
For generating the reference signal, my first idea was to use a multiplexing method. Indeed, we need to measure 8 points in the same wire with only one mesaurement equipment, so my idea was to generate one by one the reference signal at a very fast rate, and to trigger the measurement when a signal is incoming, creating a sort of scan of the screen. The other idea was to use different frequencies and make a [http://en.wikipedia.org/wiki/Fourier_transform Fourier Transform]
During the 3 first weeks, we have meetings with researchers from the university in order to take advice about the right method to choose. We have also received the development board [http://www.avrfreaks.net/index.php?module=Freaks%20Tools&func=viewItem&item_id=128 STK200 USB] so we have spent time in order to install the board, and I have increased my knowledge about WinAVR.
4th Week
We have some meetings with other researchers. And I have made the board work. We are now able to implement any C program into the micro-controller. Given that, I have create the code for generating the multiplexing signal, and it works perfectly. [http://userweb.elec.gla.ac.uk/t/tdd/ Dr Timothy Drysdale] gives us a lot of excellent advices for getting a better output signal.
5th Week
During this week, We have worked a lot on the signal and the different method in order to make it cleaner. Indeed, we need to filter the signal before any measurement cause, now, we have got a 50Hz noise. This entire week, we have worked on different filter. First, I have build a simple high pass first order filter (just a simple capacitor and a resistors. This method works quite well but we have tried to involve other filter. So, we have build second order band pass filter and second order high pass filter. But a problem was that the frequency of the signal was too speed for the filter built. Using a band pass filter was not a good idea cause it will remove the high frequency from the square wave signal and then it will deform it. Then, we have build a [http://en.wikipedia.org/wiki/Lock-in_amplifier Lock-In-Amplifier] which is an other method in order to eliminate the noise. We also have a some meetings in order to increase our knowledges on scientific aspect of our project.