in most light conditions, except for outdoor light, because of the high infrared output of the sun. Additionally, the cameras triangulated the position of the user much more efficiently than the ultrasonic system primarily because the speed of light is substantially faster than the speed of sound. Finally, whereas the sonar speakers sent signals in a more serial fashion, with the infrared system the cameras took images simultaneously. A new graphic version of the software, called Flashtrack, developed for the infrared system, was written primarily by Badger.

Around the same time as the development of the infrared system, APR received a CANARIE grant to develop a networked version of the tracking system. This networked version of the GAMS system is the current iteration in use and the focus of the rest of this paper.

The networked GAMS system is configured so that it uses a Control Centre to manage information from one remote site to another. Positional data is passed to the Control Centre and, then, onto the remote site so that a remote user can be virtually present and control media over the Internet. The positional data is passed on via UDP, an earlier network protocol than TCPIP. UDP does not include error correction and is, therefore, more effective for real-time media control. In essence, the x, y, z positions comprising the 3D grid and the speed of movement, or velocity, of the user are transmitted to the Control Centre. The Control Centre can, then, pass this positional and velocity information on to the remote site. Additionally, the Control Centre can manage several "Universes" that define who the remote users are, meaning a number of locations can utilize the system at the same time.

This version resulted in a significant shift in the way interaction could be enacted and the first instance of telepresence using this system. Now, multiple users could ostensibly interact together from different locations in their own spaces and "see" one another in the system. More importantly for this study, if one user moved her body and triggered a robotic light in her space, that movement could occur at approximately the exact time in the remote location, thus activating the robot light telepresently. (Fig. 5)

Figure 5. The networked system

Programming the space on a 3D grid with this version of the system meant that media elements could be placed in particular locations on that grid, essentially creating zones where certain sounds, images, text, or any media element could be found. Thus, when a performer enters a zone where musical notes have been programmed, for example, these notes are played as the performer passes her hand through that zone. Performers, in this way, can play notes in any way they wish, depending on the placement of the trackers. But not only can the audience hear the notes played, but this information is passed via the system over a high-speed internet connection to the remote performer in his lab. The first art piece to use the network version of tracking system was Virtual DJ by Gibson, who gave a remote, networked performance from his location in Victoria to another in Montréal in late 2002 for the CANARIE conference. Running on a research network, the system had a minimal amount of latency, around 70 ms., allowing for relatively transparent interaction. Another remote version of Virtual DJ was also performed at the NEXT 2.0 conference between the University of Victoria and Karlstad, Sweden in May 2003. The latency in this performance was slightly more noticeable, 150-200 ms., because of the distance between the two locations.

In Fall 2004 a Canada Foundation for Innovation grant awarded to Gibson made it possible for him, working along with Bauer, to research ways to create behaviors for the control of media­­such as video, animation, and images­­via the Flash Track software and the Control Centre. This work made it possible for positional information to be sent from Flash Track to the Control Centre and, then, onto a third computer running Macromedia Director and the behaviors to allow the user to match locations in the room with videos and images in Director and to manipulate these images files with user movement.

The first piece to use this technology was When Ghosts Will Die by Gibson and Dene Grigar, a performance-installation that debuted in Dallas, Texas in April 2005 as a work-in-progress. This experiment with interacting with media elements was repeated with a networked performance of Virtual DJ, originally a non-networked piece, by Gibson and Grigar in the summer 2005. For that performance Gibson, working from his studio in Victoria, BC, moved robotic lights and produced sound in Grigar's lab in Denton, Texas while she simultaneously did the same in his from her space. Documented on video, the work clearly demonstrates the notion of embodied telepresence. The addition of webcam technology, notably iSight and iChat, provided visual representation of each remote performer in each local space, enhancing the sense of performing together.

Since September 2005 recent changes to the GAMS system include tagless tracking and voice activation interaction, innovations that free the user from trackers or wearables and facilitate the use of spoken word for commanding behaviors. Bauer's Life Tastes Good, which debuted at the Collisions conference in September 2005 at Victoria, Canada, is the first art piece that utilized this version of GAMS. Because no mediating device is needed between users and sensors, the interaction is a direct relationship between human bodies and machines across space.

3 GAMS' Application for Embodied Telepresent Collaboration

As the previous section detailed, the evolution of motion tracking technology, specifically the system referred to as the Gesture and Media System, shifted interaction from that of a single user interacting with sensors and computing devices in a 2D space via ultrasonic frequencies to that of multiple users interacting with

1 2 3 4 5