When DFRobot released an inexpensive sensor with integrated face and gesture recognition for 5 distinct gesture for some reason my mind immediately went to the 5 gesture and tone sequence used as a greeting by aliens in the movie Close Encounters of the Third Kind. Throw in the 5 colors that went with the 5 tones and suddenly I really needed to re-create key elements from the movie props, especially the big light board in the final sequence. But I clearly would need to scale things down.

I decided a board of about 18 inches on a side would be the biggest that would fit in my small car along with an 8 foot stand and all the other stuff I typically take to maker showcase events.

I knew keeping the light board as light weight as possible would be critical to keeping the stand light without becoming tippy or needing heavy sandbags. Reviewing the movie it looks like they built big wooden boxes with big lights in them and put colored gels over the front. My ultralight version was to build the boxes out of black foam core board and include a white diffusion layer between the light sources and the color gels.

After a lot of thickened super glue and a lot of foam core & gel cutting the boxes came together. My big mistake in process was to allow finger prints to get on the gels before permanently gluing them down with super glue. I came back an hour later to find finger prints and smudges all over the inside of the gels like a scene from CSI. I managed to get the last one on in cleaner fashion by thoroughly cleaning it with alcohol first and wearing rubber gloves while mounting it.

For the light sources I used a series of 1 watt LED stars. These presented a challenge as they, and their resistors, could get hot yet they were glued to meltable foam core. Some aluminum flashing and extensive use of aluminum tape in between provided adequate heat barriers while staying light and even reflecting some light forward.

Electronics was generally not complicated on this since the most demanding logic (for face and gesture recognition) was already embedded in the sensor/camera combo semi-hidden in the front of the box. The electronics in the back panel mostly consist of an Arduino, an MP3 player/amp/speaker for the audio, and a bank of relays to sequence the light boxes. I tried using rotary encoders to adjust volume and recognition sensitivity but it turned out that the little Ardunio could just not keep up with the encoders. I’ll be swapping these out for something more analog in the near future.

When its up an running the unit detects the presence of a person, turns on the panel showing the 5 gestures, responds to the gestures with light and sound from the movie, and then rewards the correct sequence of gestures with the hello sequence from the movie.

Trying this out on live people at the recent Rochester Maker Faire, it became apparent that some changes are needed. It’s operation is bit obscure. People look up at it when the green gesture panel comes on, but don’t recognize that it is inviting them to make the gestures. I think I will add a new extension to that panel giving some instruction. I also need to change out the rotary encoders and maybe make it louder.