At this point, we were just making tweaks based on feedback received from our reach demo presentation.

For one, we had been working with the pillow completely deflated and taking readings from there. Rahul advised starting with a fully inflated pillow and working from there, which we did. Surprisingly, the readings from the pillow appeared to be more consistent with a firm, almost fully inflated pillow than they were when we were working with a mostly deflated pillow.

We also refined the algorithm a little bit, calibrating for each of the 4 sensors rather than by overall pressure on a side.

Overall, this was a very interesting project that could definitely make a difference in the lives of so many people. Our early struggles with getting good sensor readings, operating the motor, and integrating the three components took the majority of our time—if it wasn’t for those, we would have loved to see how far we could take the project. The wiring got to a point of near insanity towards the end, looking back it would have been nice to make a custom PCB to minimize the wire clutter. Using more sensors to get a higher resolution “image” of the pillow would have also been a great feature, especially if we could also build out more sophisticated analysis of the sensor data to inflate the pillow more precisely. The system we have built does the job, but it is quite crude. Overall though, it’s been quite the learning experience!