About

AR/fiducial recognition project I worked on for Stevenson engineering during the annual engineering trade show held in Napier, New Zealand.

We decided to build our own multitouch/AR table using the diffused illumination technique (Rear DI- http://wiki.nuigroup.com/Diffused_Illumination) for this project. Using infrared LED's wrapped around all the edges of a special acrylic surface glass, the infrared lights emit enough infrared beams that anything touching the surface glass would reflect the light and generate a strong enough image for an infrared camera to recognise. We used a hacked Sony PS3 camera with infrared lens, in the base of the table for the image recognition. To display the image on the table, we had a Hitachi ultra short throw projector also mounted inside the base of the table.

As well as the design and build of the trade show stand and interactive table, we designed and printed a range of brochures based on Stevenson's past case studies. These brochures were printed with fiducial markers on the back, allowing people to view individual case study highlights by placing the brochures on the table. We also had a few physical items with fiducial markers attached to help tell other parts of the Stevenson story. There was a sander which told the history of the company when sliding across the table, a mechanical rotary item (I don't know exactly what this was, but it was off a big cement truck) that would change the background image and play a recent news video, 2 lego pieces that when both are placed on the table would activate a game of pong/air hockey, and a few other bits and pieces.

I designed and built the application in Flash, using the TUIO library http://www.tuio.org/?flash. This allowed the application to receive messages containing data about the fiducial markers (id, position, rotation, etc) from the Reactivison application running in the background. Reactivision  (http://reactivision.sourceforge.net/) is an excellent open source computer vision framework, which handles all the object recognition and has fantastic integration with Flash, Processing, and many other applications.

 

Best part about this project, we got to keep the table and make all sorts of custom multitouch & AR apps for ourselves.