Upcycled Touch Table (Proof Of Concept)

Upcycled Touch Table

Zach's end goal for this project is to build a touch-screen table for use as an interactive map display during pen-and-paper role playing games similar to the SurfaceScapes project at Carnegie Mellon.
What we presented at the Atlanta Mini Maker Faire on 9/10/2011 was a successful Proof of Concept (POC) of the hardware side using a 17 year old (!) 42" rear project TV ($100, Craigslist), four infrared lasers with beam splitters ($70, Peau Productions), two sheets of Plexiglas ($30 each, Home Depot), a modified PlayStation Eye camera with Infrared Pass Filter ($80, Peau Productions), a sheet of Mylar ($6, Local Art Store), and about $10 worth of miscellaneous nuts and bolts.

Build Team:

  • Zach (bstrdsmkr)
  • Doug (Laz)
  • Sam (samthegiant)
  • Selena

Over the course of the last ten weeks the team has met at least one night a week after work in Doug’s garage to work on the project over five major phases.


  • Phase One: Basic woodworking - one of the first challenges for the team was simply to get the heavy old TV flipped onto it’s back and up to a height suitable for use as a table. After some basic dis-assembly we had the old frame stripped down to it’s minimum support structure and were able to us some of the existing cabinetry to add in a simple box frame of 1”x8” planks with 4”x4” posts carriage bolted to the frame for legs. 
  • (In the picture below you can see us using dowel pins to connect the framing material.)

    (In this picture you cab see the legs attached to the newly added framing with carriage bolts.)

  • Phase Two:  Infrared LEDs  Once we had an established work surface we needed to address our infrared source.  Our first concept was to illuminate the underside of the surface with infrared LEDs scavenged from an exterior infrared security light for security cameras. (See image below.) These LEDs were designed with narrow beam angles and long throw in mind and were too bright as point sources for our camera setup to tolerate and caused reflections off of the bottom of the screen that overwhelmed the signal from any interaction with the surface. At first we tried using the whole lamp, but even after pulling individual LEDs and wiring them separately they were still just too bright.
  • Phase Three: Converting the screen - The screens for these old TVs are combined from a Fresnel Lens and a Lenticular Lens, this combination optimized the image for small viewing angles but was not well suited for our purposes as a table surface. We replaced these components with two layers of Plexiglass with a layer of Mylar sandwiched between for projecting the image on.
  • Phase Four: Infrared frickin’ Lasers - Since the original LED illumination concept was not feasible on our time-scale (We were only two weeks away from the Maker Faire when we abandoned this approach.), we researched a secondary option and purchased four infrared lasers with beam spreaders to create a plane of infrared light just above the screen surface. This provided a very strong signal for the camera mounted under the surface and pointing down at the mirror. This approach did have the disadvantage of requiring that the lasers had to be aligned to the surface evenly which required modifications to the frame. (And while writing this up we learned we really need to learn to take better pictures of our build process because we have no images for this or the next phase of the project.)
  • Phase Five: Software - One item of particular importance for us to stress was that this approach relied on image processing instead of any actual touch feedback. The modified PlayStation Eye camera with Infrared Pass Filter we used was looking for the reflected infrared light from our fingers, so the image processing software running on a laptop would take input from our camera and close the loop between the "touch" and the output to the TV projector. We used the excellent Community Core Vision software. This software took the input from our camera and used a series of filters to eliminate noise, enhance signal, and then identify our "touch" interactions with the surface. (This software also would have allowed us to interface with Flash applications, but at this point in our build it was 5:30 PM on the day the before the Faire and it was time to pack-up and get moving down to Atlanta, so we did no actual software modification for this presentation.)

Trip down to Atlanta and presentation at the Mini Maker Faire - Setting up bright and early Saturday morning we learned the most important lesson of the day, if you are going to demo an infrared based system, don't try and show it off outside. We immediately split into groups, one to finish setup and the other to make a mad dash for tarps to cover and surround our display tent.

We also found tarps to be indispensable for blocking out as much sunlight as possible for accurate calibration.

We were overwhelmed by the attention that our project received.

Unfortunately, the inexpensive nature of one of our primary components became a severe drawback. (The 17 year old $100 Craigslist TV projector died about four hours into the Faire.) We gave it a most glorious sendoff into a campus dumpster right before heading home.

Lessons Learned:
We learned a lot during this build and presentation including:

  • Presenting an infrared detection based system outside is a bad idea.  Even under a tent covered with and surrounded by reflective tarps we experienced an amazing number of false signals that made everything other than our simplest calibration software go hay-wire.
  • We need to make sure to post at least some build note specifics before leaving for a major presentation like this. (I would estimate that the number of people who took snapshots of our dinky little sign or asked about a URL would easily be in the low three digits.)
  • We need a Knox Makers sign. (A large one, at least 3’x5’)
  • We need a Knox Makers logo
  • We need Knox Makers business cards and/or flyers to hand out with our information
  • We need to prepare a FAQ board and handout sheets for build presentations (How much did the build cost, what components and software were used....)