Uploading

The cocktail dress as worn for the the readings. The accelerometer is tucked in the hair with the wires coming out in what looks like a loose strand of hair and running to the chest where the rest of the electronics are housed. Measurements as followed: nodding (x-axis, 400-460/down-up), shaking head (y-axis, 570-500/right-left)

The good news: the dress works. At least, the dress works in terms of taking in data, broadcasting it to the computer, and writing it to a data file. The problem is, while the measurements are accurate, they do not not accomplish the exact task I set for them.

Ranging

My concept of the cocktail relied on a quite specific situation: people standing up at a party and talking directly to each other. I positioned the proximity sensor on the chest so that I would be capturing the larger targets of the body rather than the smaller targets of the head. However, after experimenting with the sensor, the sensitivity is great enough that this concern for the size of the object is unnecessary.

Furthermore, it means that the dress is limited in value when in other situations. Since I missed the intended window for testing, I wound up having to use it in a suboptimal context: sitting on a couch with friends. While we did talk and I nodded, shook, and threw my head back to laugh with the best of them, I was turning my head, not my torso to talk to them. This meant that, while I was appropriate values from the accelerometer, I was registering the proximity values of my computer in front of me instead of the people I was talking to.

Since I have found that it is not necessary to be taking readings from the torso, the clear solution is to mount the range finder on my head with the accelerometer.

Accelerometer

Actually, the accelerometer worked very well. It tucked neatly in my hair. It seems like a small, head mounted electronic is certainly the direction to take this project, whether as a hair piece or a hat.

Context

Another problem that hadn’t even occurred to me was setting the context of these readings. The proximity sensor gives a very limited sense of context for the accelerometer readings. It is also a reading that would gain value if I knew I was talking to a person and could then use it to detect how close the person was to me. It would be very benificial to also be taking audio so that I can hear what it is I am reacting to. Am I laughing at a joke? Am I agreeing with some statement? Am I positively rolling on the floor? Hopefully, I will be able to integrate a microphone with my next version.

The unassembled components of the cocktail dress

Next Step

Since the current set of readings are of limited value, I plan remake the the piece as a hat or hair piece and retake the values at ITP’s 30 Anniversary celebration. The environment is much more along the lines that the piece was meant for and this improved version should take much better values. In addition, I hope to borrow one of the ITP phones to log the values and, hopefully, I can take audio and record it too.