CESTUS

During the week Tom has written a Matlab code script that can detect which sign we are making with the glove in theory. We tested the code with the glove. The code worked but was fairly inaccurate. We then recorded some more values to try and make the program more accurate

We have began recording the different sign language gestures and storing the results in tables. At first one of the flex sensors was giving irregular readings because the contacts were touching another flex sensor and shorting it. The sensors were repositioned slightly.

We tried reducing the significant figures but it made hardly any difference. We noticed that the noise level in our circuit had decreased from the previous week. We think that the noise might have been caused by another project with our long wires acting as antennas. We have marked where the flex sensors will go on the glove with a marker pen, we have attached all of the flex sensors to the appropriate positions on the glove using super glue. We have attached the circuit board to the glove using a hot glue gun.

We are coming to the conclusion of this 5-week project, today we shall be glueing the flex sensors to the gloves and recording results from the gesture movements of this flex sensors. Everything seems to work perfectly so far already, we've tested the codes and Matlab codes to ensure that everything works fine and seen the plots.

Jonathan Sims , in charge of this design drew the placement allocations for the flex sensors in order to help when we decide to glue the flex sensors to the gloves. This was done using visible markers on the gloves. There was a test glove where we glued some of the flex sensors to test the effect and the flexibility of the glue when we do decide to do this on the real gloves. Glove Color: Grey and Black

We found that there was some pulsing (output voltage was fluctuating up and down) in one of our ground rails we replaced the ground wire.

We started by removing any shorts on the circuit board using a soldering iron. We found the the total resistance of the circuit was only around 5V which indicated that there was still a short. We found that some of the tracks on the circuit board were not cut correctly. We cut the tracks and found that the resistance of the circuit had increased to 18(kiloohms). We connected the circuit to the micrometer again and it didn't crash. We did however get some invalid readings of 0V on some of the outputs. We noticed that the variable resistor was not connected to a positive voltage supply so we soldered on a wire that connected the positive voltage rail to the variable resistor. The outputs no longer read 0V but they were fluctuating between 0V and 0.99V.

You can tell it was difficult to get this program to work! Implemented a system of semicolons to signify new cycles and commas to signify new values. There are "\n" after each number to terminate the "fscanf"

This took a really long time to get done because of the need to be accurate and make sure to reduce the possibilities of errors that can be encountered

We designed the Analog-Digital Converters to save the number of wires that will be on the board  We connected a potentiometer to set the maximum voltage to the Voltage reference(V_ref) to get the maximum range of values from the flex sensor.  This is the variable layout design for the circuit with components such as op-amps, resistors and flex sensors.

After the first week session, we were able to build a circuit we would use for the smart glove, we also worked backwards to obtain a suitable diagram and wrote a code using "arm Mbed" to program the code for the microcontroller. We used a serial monitor to monitor the movements made with the flex sensors and our results were impressive.  We also got down to soldering some of the components on the circuit board.