Last summer Ray Chi got in touch with me about an installation he was doing for the Milwaukee Art Museum. He wanted a video screen that could be activated to play a video by touching a metal plate. Well, actually six videos and six video screens and six metal plates.
I told him I’d figure out how to get it all to work the way he wanted. I came up with a few ideas, one of which was using Processing, which I did get working, but at the time Processing wasn’t really running on the Raspberry Pi, at least not officially (or very well) and since the Pi was what we ended up choosing, I needed another solution.
For an installation that’s going to be running for years, simplicity and reliability are key. I had used Pis in the past many times for video players using omxplayer. Typically I’d just launch omxplayer on boot and have it play a video, looping, forever. For this application we wanted the video to play only when the metal plate was touched by a human hand (and then stop playing when someone stopped touching it) which meant capacitive touch.
Rather than spend a lot of time coming up with something that might work, I went with something that I was 98% sure would work. I used Adafruit’s Standalone Momentary Capacitive Touch Sensor Breakout attached to a Teensy LC. Why a Teensy LC? Because it’s a low-cost (LC!) Arduino-compatible microcontroller and it can emulate a keyboard.
Yes, a keyboard! If you have a USB keyboard connected to a Raspberry Pi computer while omxplayer is playing a video, you can just hit the space bar to play the video, and then hit it again to pause the video. Those are the two things we needed to do.
So, Raspberry Pi, running omxplayer to play the video, with a Teensy LC attached programmed to work as a USB keyboard, and triggered by a capacitive touch sensor, which was then connected to the metal plate. Simple!
There was this issue of……. timing.
So in theory, the Teensy would just need to send a space character to play the video, and it would do this when you touched the metal plate. But! (And it’s a Big But) the issue was that we wanted the video to start playing at boot and then pause at the beginning and sit there waiting… for someone to touch the metal to start the video playing. Rather than fire up the video via the typical Linux methods, we ended up just starting up the Pis, auto-logging in, and having them wait at the command line… yes, just sit their waiting, doing nothing… Sort of.
When the Pi booted up, it provided power to the Teensy, which then started running its sketch. The sketch would start at boot, wait 45 seconds to ensure the Pi was booted up and sitting there waiting at the command line, and then it would type:
/bin/bash /boot/video.sh
So we actually used the Teensy to send the text to the Pi (just as if a human typed it) which then fired up the script and started the video playing. The sketch would then wait 2.4 seconds and type a space character, which would pause the video. This set the state of things exactly where we wanted them. The video way paused, just waiting for the next command from the Teensy, which was… space, of course!
Now, there’s the concept of “rising edge” and “falling edge” when it comes to pressing buttons. A rising edge is the transition from low to high, and a falling edge is the transition from high to low. That’s a fancy way of saying we can tell when the button is being pressed, and when it’s being released. It’s best to use debouncing for this, and there’s a library for that.
So with everything mounted in place we still had to deal with one issue. The HDMI displays worked find as long as they were turned on before the Raspberry Pi computers. If they were turned on at the same time the resolution wouldn’t set right, and the video would be letter-boxed. There were two options, one would be using two different power strips to get power to everything, with instructions for museum staff to follow a specific order. This wasn’t ideal, so we went with option two. I used a time delay relay so that one single power strip could be turned on, which would turn on the HDMI displays, and then a few seconds later turn on the computers. It worked. (And yes, I found out later I probably could have fixed the issue in software. Duly noted.)
Are there things we could have done better? Yes. Did we get the project done on time, and within (or under) budget? Yes. Was it fun and challenging? Yes and Yes. You may read this and think “Hey, you totally could have solved problem X by doing Y!” and you’d probably be right. I’ve found a number of things I’d do slightly differently if I were to do something like this again. That’s all part of experience, and learning, and sharing… right?
When the installation was all done and tested, I got photos of everything, and then set to work on documenting it all. I delivered a 14 page manual on the construction and operation of the video players, along with the code and instructions on how to use one of the backup SD cards that was prepared in case of failure.
Besides, now I can (sort of) say that my work is in the Milwaukee Art Museum. ;)