Our approach to interactivity, which we call “environmental,” creates a responsive virtual world attuned to the life of the terminal building. Since Horizon would not direct itself to specific individuals but rather to everyone in the public space, its interaction is with the global conditions of that space and with the over-all behaviors of the crowd rather than with individual actions.

Thus its responses are never compulsory. Monitoring its own levels of activity, variety, and even boredom — it would choose when and with what to interact depending on its internal state.

That said, interactivity would be crucial to Horizon, for its continual insertion of unpredictable elements would insure that the display continued to change — and even to evolve, as the artificial intelligence program would begin to recognize and build upon patterns detected over time in the environment.

 

Inputs

 

Calendar

The computer knows the date and time from its internal clock. These simple calendar functions yield surprisingly rich results: not only do they give us the season, the day of the week, and the nearest holiday, but they also tell us the exact times of sunrise and sunset, and even the exact slant of sunlight at any given moment.

Augmenting this internal information are three video cameras trained on the Departure Hall, which enable two further inputs:

 

Light

The cameras can “see” the quality of light both inside and out, telling us when the sun is blocked by clouds or when it is casting strong shadows within the terminal.

The quality and degree of light affects people deeply, if for the most part unconsciously. This is especially true for jet-lagged travelers. One of the simplest pleasures we can offer, therefore, is bright light on a cloudy day, as Horizon becomes a respite for weary passengers stumbling off an arriving flight.

 

Crowd patterns

The richest and most changeable set of patterns in the building is that of the terminal crowd itself. “Crowd” implies a singular group, but in fact the Departure Hall and the terminal as a whole will contain numerous sub-groups that are continually forming, breaking up, and reconfiguring.

The pattern recognition software analyzes the video footage in real-time to detect the:

• Massing of figures, from which may be deduced sub-groups and individuals in the terminal space
• Relative density of figures within such groups
• Shapes of groups (lined up, bunched, evenly or unevenly spaced, dispersed)
• Speed of groups (standing, walking, running)
• Direction of groups, whose trajectories can thus be traced
• Over-all variability of motion (both within and between subgroups)