And this is the end...

First, as usual, some pictures. Being rather shortsighted, we didn't take a lot of videos, but we have tons--dozens--of pictures. Below are a few highlights:

The final circuitry. We didn't show this on Demo Day because the pressure data was accurate but not that interesting, and our GUI was only programmed for a simplified output for our testing purposes- i.e. whether there was no/low/high pressure applied. The accelerometer data gave cool-looking results, though, and since we only had less than 1.5 minutes for the demo portion of our presentation (we found we couldn't summarize the technical stuff quickly enough, there was so much to explain :( gah, sorry!!!), we figured we could just wire up the accelerometers on a separate breadboard and demo with them. From experience, we tend to greatly respect Murphy's Law- in other words, the more circuit you have, the more things can go catastrophically wrong........... and on the night before Demo Day, we had a lot of weird circuit issues which might have been related to the pressure sensors, and our primary concern was avoiding showing a completely non-functional device.

The completed inserts! The sensors are located in the areas corresponding to the big toes, the balls of the feet, and the ball of the heel. We foolishly forgot to take a picture of the actual wiring of the inserts before I sewed them shut, but the ribbon wire just branches out (2 wires per sensor- 1 to the pin, 1 to ground). The final thickness of each insert was 8 mm!

The accelerometer on the toe. 

We made this insert size 10 so that Winslow could wear it :)

...and it fits perfectly!

HOW IT WORKS

The gait sensor measures the orientation of each foot as well as the pressure distribution across the sole of each foot as a function of time. The physician prescribes the use of these custom shoe inserts for a set length of time or during a certain activity—e.g. throughout the course of one full day.

Once the insert is placed in their shoe, the patient switches the connected device on, wears the wire-connected device on a belt clip, and lets it record data throughout the time prescribed as they go about their regular activities. Upon completion of the prescribed time interval or activity, the patient switches the device off and brings the insert with them to the next office visit. The physician can then remove the SD card enclosed in the device and read the data into the provided graphical user interface (GUI) to observe the patient’s gait pattern.

This is is a tiny fragment of testing where I demonstrated the working pressure sensors to Winslow via iPhone video while working on this outside of class... we also foolishly forgot to take a more formal demo video before we dismantled things to return the parts. Sorry, that was just lack of thinking on our part :( Also, on the day before demo we could only get the sensors working occasionally, and to this day we don't know why. The pressure sensors worked well when they did decide to work, though... and they also happened to work when I tried them after demos were done. That figures. Anyway, there was a nice linear response to pressure applied, as the data sheet from Flexpoint promised, when I tried having a couple friends (of known mass, haha) stand on the sensors.

This is the demo of the MATLAB GUI that Winslow made... the time reads in seconds here, and in future iterations it will play back the data at a rate that the user specifies (so it can play as fast/slow as desired). The dots corresponded to no/low/high pressure applied at the time of this demo shot.

The final material cost per insert (including base, outer covering, thread, bend sensors, wiring, and accelerometers) came out to roughly $17. We recognize that there are a multitude of other costs, both direct and indirect, to consider (the microcontroller or other hardware that will ultimately be used, testing, manufacturing, labor, approval by appropriate panels and authorities, etc.), and we also need to consider the additional cost of ramping up the number of sensors per insole. However, when we consider that cost with scale-up in mind, we think the relatively low material cost is a promising indicator of the device's potential. Certainly we don't anticipate this device reaching the $4,995++ price tags on competing devices and systems.

Things we could have done better:

• We could have requested additional samples from Flexpoint to get a better idea of the final hardware , but we decided not to due to time constraints and, well, general courtesy.
• The GUI could have had more features, but its development was also constrained by time.
• We should have taken more videos as we developed and built our prototype.
• We should have done more research on what material would be best for the insert base. (Thinner, less intrusive, lighter, allow wireless transmission)
• The device could and should have been smaller! Maybe with a smaller battery pack and a different SD shield/breakout board.
• We could have built a protective case to house the circuit boards and batteries. 
• We could have created additional feedback systems for informing the user that the batteries are low or if the memory space is running out.
• Ultimately, the bend sensors will also constrain our design: once you make them small enough and pack them close enough together, the bend sensors won't actually all bend in a region that is compressed--at least, not in the ways we expect. We think this constraint will not have a bearing on the final design of our product, however, because we are not aiming for the level of resolution offered by the competitors' products. A lower resolution would be sufficient to observe the gait patterns that physicians are really interested in. (The exact minimum density to observe these patterns using our sensors of choice is unknown, though, and requires some additional testing.)
• We would like to develop this further so that data transmittal is wireless--whether it be via an online method so that the physician receives the data directly from the device (eliminating the need for physical data storage) or just wirelessly to the recording unit worn on the belt clip. This would eliminate the wires that currently lead from the inserts to the Arduino and make the inserts almost completely unnoticeable!

All in all, we had a lot of fun learning how to work with the Arduino and developing the different aspects of our device, all the way from initial research to physical realization! It was a crash course in everything from Arduino programming to product design for me, and I learned a lot from Winslow, too.

I was personally really inspired throughout the semester as I looked up things that other people were doing with this microcontroller, both in and out of this class--to the point where I'm currently considering buying one for my younger brother to play around with as well. It's an amazing way to get first-hand experience in programming and physical computing, and both of us are glad we got the chance to take this class! Much, much, MUCH thanks to Dr. Bogen for overseeing our work and teaching this course!