To be able to test the serial reader, I also wrote a serial writer so that I could control the reading and writing of data across the port.  I was able to compile and open port “COM1” on my computer for read and write.  To test my read and write serial programs I also had to install the javax.comm package on another computer so I could link them with a null modem cable.  Due to the fact that the other computers in my lab area did not have jdk1.5, Jason and I downloaded javax.comm on the two computers in his lab, but unfortunately when the code was ran it would terminate before the port was opened for reading and writing.  I then tried implementing the programs on my computer and one that I had installed javax.comm on, which resulted in a termination also.  I then figured out that I was not correctly closing the ports for them to be reopened when the code was run again.  After fixing this, I could open and the close the ports every time each program was run.  

This week, I also reviewed the code for the Motorola chip that was used for the SEMGO project.  I discussed with Jason what would need to be changed to successfully send data across the serial port to be used to manipulate the mouse.  For most of the week, we were unable to do any work with this aspect of the project due to the fact that we could not compile the prewritten code and download it onto the chip.  Once this was resolved we added in the functionality to send the data that was collected from the EMG sensors across the serial port instead of remotely controlling the robot.  Before we were able to test our code a program was written to send data through the serial port and read using an existing C program. I was able to conduct some research on active sensors.  Barani supplied me with some great internet sites that had multiple active sensors and from that I was able to find some other ones.  

This past week I finished the code to read from the serial port.  I was able to read a series of numbers written to the port by our board.  Two other programs, one written by Caleb Wells a fellow REU student for another project and a built in HyperTerminal program, verified that the data being read was accurate. 

After testing my reader I concluded that there is a problem reading in first number in the sequence. Jason’s test program for the board writes one number continually onto the serial port and when my code is run the first number read is always different then the following numbers.  The port blocking some of the bits at the beginning may cause this problem, but either way this problem will have to be fixed.

To accurately read from the serial port, the port needs to be closed after each time the code is run.  If the port is not closed then it may not properly open the next time the code is executed.  To insure that the port closed I needed to find a way to break the loop that was reading from the serial port when the user has ended the program. 

Jason and I thought that a way to break the loop would be to have the user press a key on the keyboard.  Through the power of Google and java.sun.com I researched how this could be performed.  After looking at keyboard event, key listener, key_pressed functionality, and programming test code, I realized, with the guidance of fellow REU student Brian, that without a container or GUI I would not easily be able to implement what I had wanted to do.

I implemented a small GUI that allows the user to stop reading the input stream and close the port.  The GUI also allows the user to reset, which will close the port and then reopen it to read the input stream.  These buttons can be pressed by the mouse or through keyboard commands.

With the data being read in at such a high volume my GUI became very slow.  This problem may be fixed through the implementation of threads, but may not be needed because data being read from the EMG sensors may not be written at such a high rate.

This week I researched different muscle groups to use to move the mouse cursor.  Dr. Gutierrez suggested that I look into the shoulders, neck (rotation, pitch), and face muscles.  These muscles could potentially be available for use on a person who is not able to move a cursor due to immobility of their arms or hands.  I found it very hard to identify different muscles within the neck that a specific movement correlated with and in turn did not also contract many other muscles.  I looked at rotation, lateral flexation, hyperextention, and flexation within the neck.  I also looked at the muscles responsible for shrugging the shoulders (trapezius –upper), clenching the jaw (masseter), and raising the eyebrows (occipitofrontalis).  To test if these muscles may be used for the project, Jason placed two sensors on his face, one on the masseter and the other on one of the occipitofrontalis (forehead) to review the strength of their signals.  Their signal overlapped slightly but overall we believe that they could potentially be used for someone who only has muscle movement in their face.  We also spent part of the week trying to figure out the problems with the hardware so that we could calibrate it with just the forearm muscles, but we were unsuccessful. 

This week Jason and I did final testing for the demo of the project due on Wednesday.  After rewiring of the part of the board we were able to use the muscles in the forearm to move the cursor up and left!!!  We were able to successfully demo this to the lab on Wednesday.   At our lab meeting I also presented my research concerning the different muscles that could be used to control the mouse in the shoulders, neck, and face. 

This past week I continued with my work concerning the different populations that may benefit from our physiological cursor. 

Through this research I realized that the benefit of the physiological cursor is its adaptability for different populations.  With each population, our physiological cursor will be able to adapt to the specific muscle groups that are not affected by the specific disorder and will allow the user to use those muscles to manipulate the mouse cursor on a computer screen.

I completed research on alternative mouse products, to compare their functionality to the physiological cursor.  These products included: eye tracking, head tracking, and voice recognition.  In each case these products needed specific muscles groups to be functioning to work, where the physiological cursor can use any of the muscles that are not affected my the disorder to manipulate a mouse.

For proof of functionality, I started on reworking my code so that the commands it receives are the same as those sent by a serial mouse.  This would then show that indeed our physiological cursor has the same functionality as standard mouse. 

A serial mouse sends packets of three bytes that tells the computer the length of the x and y movements, and right and left click and release.  The program that I had working just took one byte as its input and would move the cursor accordingly.  I first started hacking my code a part to try and make the serial port functionally to work, but in the end I decided to start a new project to insure all the functionality worked properly. 

This week I completed the revisions on the physiological cursor software which allows the cursor to move on input from a serial mouse.  This task required me to change a lot of how I had set up the original program.  A Serial mouse sends a series of serial packets consisting of three bytes.  Each byte consisted of 8 bits which when combined with each other would give the x increment and y increment of the next mouse movement and a 0 or 1 value for the left and right button.  In complete serial packet the first byte would have the 6th bit as a 1, and the second and third bytes would have their 6th bit as a zero.

To ensure that each byte was being processed, I added a serial port listener and all data that was received was placed into a vector.  The first byte in the vector was checked to see if it was the beginning byte and if it was that byte and the following two where then read and decoded into the x increment, y increment, left click, and right click values.  These values where then used by the java Robot class to manipulate the cursor on the screen.  The cursor still began in the middle of the screen and all movements were still checked against the dimensions of the screen. The GUI was also recreated using threads and consisted of three buttons for user control of the program: Start, Reset, Stop.

The project was demoed for Dr. Gutierrez on Friday.  It was demonstrated that the serial mouse could perform the same functionality as a standard mouse, including actions such as drag-drop, double click, and highlight.  The hardware was also revised to send serial packets to the software depending on the signal it received.

  This week I also had the opportunity to tour the Texas A&M Low Speed Wind Tunnel.  The students that lead the tour were studying Aerospace Engineering.  We toured the machine shop where exact models were built for the tunnel.  We also saw pictures of Lance Armstrong testing his bike in the tunnel and Harley Davison Bikes being tested.

At the end of the previous week, Jason and I performed testing concerning the strength of the signals that a small muscle produced and that of a larger muscle, the forearm, and we found that the smaller muscle’s signal was around 1/6 of the larger muscle.  During our weekly meeting on Monday Dr. Gutierrez discussed different options with Jason and I concerning how to more precise read the muscle signals on the smaller muscles, like those on the neck.  We discussed the option of using a notch filter on the signal before it was processed in the Motorola chip.  A notch filter would pass the entire signal minus the 60-hertz power signal.  With this filtered signal the chip would be able to distinguish between smaller signals. 

Jason and I constructed the notch filter on a prototyping board and after much tweaking and testing (it was quite a long process) we completed the filter.   The filter did filter out noise at 60Hz caused by AC power.  It did then keep the noise at 60Hz from dominating the signal, and increased the signal to noise ratio.

This past week I also started the introduction and background for my final paper.

  Week 10- August 1^st -5^th

This week has been my last working week and it was spent finishing my poster presentation, final paper, and website. 

Jason and I also spent a lot of the week trying to collect data concerning different calibration techniques.  These were contracting the muscle hard vs multiple soft contractions.  We were able to read two different channels of data concerning the decisions that the hardware was making, but after many hours of deliberation we could not interpret the graphs the FFT data collection.  The problem was is the person who wrote the original code for the SEMGo project was only available to communicate over email which made the data collection process quite tedious and not very productive.   

The brown bag lunch was: "Evaluations" with Beth Johnson.  I thought this session was important to insure the success of the program for the next year.