Environment for the Interactive Design of Emergent Art

John D. Mitchell and Robb E. Lovell
Institute for Studies in the Arts
Arizona State University
Box 872102, Tempe, AZ 85287-2102
Tel: (602) 965-2709
Fax: (602) 965-0961
email iejdm@asuvm.inre.asu.edu


The EIDEA project grew out of movement sensing research at Arizona State University's Institute for Studies in the Arts. Described is a system that combines the emergent phenomena of artificial life combined with real-time local weather input to create a self propelling music and visual installation.

There are three parts of the system which operate in a real-time generative capacity: the artificial life world, real world weather, and chaotic functions. The artificial life world simulates the interactions of wolves, flocking birds and trees through the use of genetics, L-systems and motion algorithms. Real world weather is sampled through the use of a computer-linked weather station which inputs temperature, wind speed and direction, barometric pressure and relative humidity. The chaotic functions used are similar in form to those describing predator-prey interactions and many other natural chaotic systems.

These elements are linked to provide the viewer with an auditory and visual experience. A quadraphonic sound environment provides a direct link between the movement of the life forms in the artificial world and performance of the 48 tone equal temperament score. Score generation is also linked to the other elements of the system. A large screen projected mural is displayed which represents a histogram showing the life span of the creatures of the artificial world. Because the mural is influenced by interactions of the artificial world it is also influenced by the outside weather conditions.

This work will pay particular attention to the generative systems utilized including genetic and chaotic algorithms, how the 48-tone equal temperament score is generated, and the nature of the real time parameter modifications affected by the weather data.


Robb Lovell is cultivating a garden like no other. It does not contain the typical roses hibiscus, tomatoes, or sweet basil, nor the usual insects. Instead there are three distinct types of life carefully designed to insure long term coexistence. Lovell's Garden is a cyber-reality comprising the core of the EIDEA project (Environment for the Interactive Design of Emergent Art).

The first installation of EIDEA, created by computer scientist/dancer Lovell and composer John D. Mitchell took place this summer at Deep Creek School outside Telluride Colorado, as part of Arizona State University School of Art's summer program. Dan Collins, director of Deep Creek School, states that, "Part of the reason for having Mitchell and Lovell as visiting artists was to see what they could come up with as far as interacting with the natural environment of Deep Creek." EIDEA is a direct response to this challenge. In this work Mitchell and Lovell drew on their experience in creating performer activated stage environments and created an installation based on input gathered from natural phenomena. This input is modified and drives an artificial world installation.

The artists have been creating movement-activated stage environments for the past ten years. Lovell began working with the "EYES" system at ASU in 1988, while Mitchell, with choreographer Gary Lund, began the Movement Initiated Sound Events projects at the University of South Florida in 1985. Together they have created new systems with greatly expanded performance possibilities at Arizona State University's Institute for Studies in the Arts.[1] Collins notes, "While their work stems from a desire to advance sensing systems in general they are involved in virtual theater environments that enable a performer to influence music, video, lighting, and other theater elements from the stage space. Ultimately, they are interested in giving artists tools to create new metaphors. Their work inverts the paradigm of a performer following a set script, choreography, or program and creates a situation in which a performer initiates events. The stage environment becomes a highly responsive instrument with which to perform. The emergent world idea came from their work in controlling the graphic output of the sensing system. They began to understand that this graphic output was, in and of itself, an ongoing dynamic event," leading to the creation of a work featuring real-time graphic generation.


" EIDEA" according to Lovell, "exists as a mathematical plane floating in a vast space. Three artificial entities exist on this planeÐbirds, wolves, and treesÐand interact to form a closed ecosystem." Trees, formed by an L-system [2] create a fractal shape, reach maturity in ten years (y-units), bear "fruit" and live to an expected 100 years. Birds consume the fruit of the trees, behave in a flocking fashion, breed and create offspring that then join the flock. The cyberspace predator, a wolf-like creature called a turoid (after computer scientist Allen Turing, feeds on the birds and has its own breeding and life cycle. Trees grow slowly, some 100 times slower than the other creatures, and over time bear fruit for the birds and wolves to eat."

Artificial life is a study into the inner workings of nature through the use of technology. The earliest tool technologies allowed man to manipulate the world around him to alter the natural order to suit his purposes. However, some things about nature can't be modified, only tested, observed, and modeled, allowing predictions to be made about the outcomes of particular events such as the advent of floods, or the change of seasons. In the past, artists used simple technologies to recreate nature, using music, painting, dance, and sculpture to capture the static or semi-static forms of living things. Early technology consisting of pneumatic devices including floats, siphons, and the waterwheel, were used by the early Egyptians in Alexandria to model time and create gadgets in the shape of animals. Later, with the invention of the mechanical escapement and pendulum, artifacts consisting of complex behaviors allowed for a more precise modeling of time. As technology improved, man's models of nature progressed and became more complicated. Over time more and more complicated mechanical systems were devised, such as levers which converted circular motion of a cam into linear motions. This provided the means for the creation of complicated mechanical automata which looked and acted like real animals or humans. An example of such an animal is Jacques de Vaucanson's duck, circa 1735, which was described as "an artificial duck made of gilded copper who drinks, eats, quacks, splashes about on the water, and digests his food like a living duck". One wing of the duck contained over 400 articulated pieces.

Finally, with the invention of the multipurpose modeler, the computer, and the formulation of the notion of an algorithm being the logic underlying a model, regardless of the model's physical manifestation, (Church, Kleen, Gödel, Turing, and Post), allowed the blooming of the modern incarnation of Artificial Life. John von Neumann was on of the first pioneers to formulate a computational approach to the generation of lifelike behaviors. His idea was to formulate an automaton capable of reproducing itself and he proved that machines could be formulated with the capability of self-reproduction [3]. Many other experiments have been carried out since that time which recreate elements of life as computer models. Cellular automata, L-systems, and genetic automata are some of the technological tools of the artificial life modeler. This tremendous advancement in the technology used to observe nature has given the artist the capability to not only represent nature in static states, but to recreate it dynamically.

In EIDEA, life forms breed through a process of natural selection. Each form has a genetic make up which determines how well it survives in the EIDEA environment. A set of behavior genes are assigned values to determine how well or to what extent each animal can accomplish living tasks. Some genes include: intelligence, strength, comeliness, dexterity, perception, and constitution. The more fit animals are selected for mating, and their genes are combined, and passed on to future generations. Less fit animals tend to be unable to breed and die off because they do not have the tools to survive. Gene mutation is used as part of the genetic model in order for the world to maintain adequate genetic diversity, without it, all the animals would tend toward a single genetic makeup.

The behavior or movement of the life forms is modeled to give them unique characteristics. "Animal motions are determined through two types of algorithms; one for bird motions, and one for wolf motion" says Lovell. "Birds behave according to a flocking algorithm. Each bird is bred to avoid predators (wolves), to stay within the boundaries of the world, and fly close to the center of the group of birds immediately closest to it." The birds are not given any instructions about where to fly, but rather are given ways to behave. These behaviors in turn cause the birds to collectively move about the world in a flock. The algorithm used is similar to an algorithm invented by Craig Reynolds in 1989 that produces flocking type motions.[4] Specifically each bird keeps track of where two to five birds in its immediate area are located. The number of birds tracked depends upon the birds perception gene. If a bird falls behind the center of its group of birds it accelerates, if it gets ahead it decelerates. The bird is always moving toward the center of the group. Should some external object come to the birds attention it modifies its motions by including a desire to move away (or toward, depending on whether the object is food) the external object. The two motion vectors, flocking and external object avoidance are added together to produce a composite motion."

Wolves, on the other hand, "are much more independent, moving alone and in a pattern specified by their own genetic code. Specifically, they move according to an inherited Turing program, which consists of a list of motions and an ordering of those motions." In 1936 the British mathematician Alen Turing created a theory describing the simplest type of computer [5]. In the theory, this computer although simple in construction was able to produce any computation. The machine consists of a tape of memory cells and a processing unit called a head which can move up and down the tape storing or erasing a 0 or 1 on the tape according to a list of instructions called a program. A turing program consists of a finite list of quintuples of the form <current state, read color, write color, new state, move direction>. Eidea uses this same concept for the motions of the wolf like life forms. Instead of a tape, a plane is used for the storage device and instead of only forward and backward motions, turns are allowed. In addition, the number of states possible is expanded from 0 and 1 to hunt, sleep, eat, mate, and explore. Each wolf has part of its genetic code dedicated to describing the possible motions the wolf can use in different situations. As wolves breed over time, their strategies for surviving increase as more fit algorithms are combined.


This type of artificial reality usually exists completely within the confines of the computer, but in EIDEA the cyber-world is actually influenced by the outside world. A weather station links the computer-generated life forms to local temperature, wind speed, wind velocity, barometric pressure and relative humidity. In the same way that cosmic forces influence the weather of the earth, and eventually our own day to day existence, local weather has an impact on the behavior of the cyber-entities and on the development or evolution of their world. According to Lovell, "The weather of our outside world influences the behaviors and abilities of the creatures in the artificial world. Wind velocity causes the birds to have some trouble flying, actually blowing them around at times. Creatures hunt and eat more during warm weather. More breeding occurs when the weather is cold."

This artificial ecosystem is also capable of exerting some influence on our personal environment. Within the context of the EIDEA installation, the artificial world interacts with a soundscape and a visual mural, providing an ongoing record of cyber-activities. A Quadra 800 generates Mitchell's soundscape by using Max and two SampleCell cards. A serial link to the artificial world housed within the Indigo II provides Mitchell with statistical data about the artificial world as well as information about the current state of the artificial life forms. This information is then used to modify score parameters such as the number and type of instruments playing at a given moment, note velocity information, and overall loudness. The Quadra 800, linked directly to the weather station, processes the weather information and passes it on to the Indigo II. The remainder of the score is based on current weather information including temperature, relative humidity, barometric pressure, wind speed and wind direction. The quadraphonic soundscape, completely generated from a combination of external weather data and information from the cyber-world, becomes an intermediary or crossing point between cyber reality and the natural or outside world. Thus the viewers place themselves at this nexus upon entering the installation.

An electronic mural, created as a history of the creatures' movements, completes the visual portion of the installation. "Constructed as a record of the motions of the animals math space, the mural results from recording the paths of the life forms and then sampling a set of images based upon their genetic makeup. These colors are then recorded in the mural for display," says Lovell. These images are projected as they manifest becoming a visual element in the installation and providing another link between the viewer and cyber reality.


Figure 1. System Overview

EIDEA is designed to be viewed in a small enclosed room approximately fifteen feet square. The equipment needed to run the installation includes two computers (one SGI Indigo II XL and one Quadra 800), an audio rack [6], at least one large screen projector, at least four audio speakers and two audio amplifiers. The work's duration can be a minimum of several hours to a maximum of several weeks.

Once the world is initiated, the score is self generating and will continue to run until the world is ended. Sound sources include two SampleCell cards in the Quadra, an outboard analog synthesizer module, and a microphone. Sound generation and control are based on data obtained in real-time from two sources; a Davis Weather Monitor II and the artificial world running on the Indigo II.

Opcode's Max is used to implement the serial link protocol for the Weather Monitor II, providing the Quadra 800 with a continuos stream of weather data as shown below. This data is then processed and mapped to the sound score in a variety of ways ranging from
directly controlling the audio mix to manipulating sound generation parameters.

Temperature is mapped to a drone that repeats in a fixed rhythm. The period of the repeat and the pitch are increased as the temperature rises.

Barometric pressure is thresholded to .01 inch of mercury, and then is tested for change. A percussive sound (generated from SampleCell) is linked to this test. If the barometer has changed more than the increment set, (usually about .05 inches of mercury), then a sound will result. This threshold can be easily changed for instances where the barometric pressure fluctuates, such as during storm conditions.

Relative humidity is mapped directly to a MIDI controlled mixer to adjust the sound level of Deep Creek. A microphone is placed just above water level where the creek flows over some rocks, creating a natural example of chaos at work. This sound becomes more prevalent in the installation as the relative humidity rises, which is usually at night.

Wind speed and direction are used to control elements of pitch generation algorithms based on chaotic functions. These 48 tone/octave equal temperament pitch generators are mapped into two distinct groups of instruments; one based on bowed string samples, and the other based on double reed samples. For the string samples, wind direction (0 to 360 degrees) effects overall pitch set size, ranging between one and nine octaves. Wind speed inversely effects the rate of note generation, with a steady state generating approximately one note every 60 milliseconds. Note duration is linked to the rate of note generation. Gusts of wind, then produce a slowing of an otherwise murmuring type of soundscape, drawing momentary attention to what has previously been perceived as an ambient texture. Finally the string samples are mapped to the flock of birds as it moves in 3D space, mirroring the birds motion in the quadraphonic sound space.

The double reed sample is linked to statistical data from the artificial world, and pitches are generated in a fashion similar to that of the bowed strings. The size of the reed ensemble is linked to the number of colonies present in the artificial world.

As colonies die out the size of the ensemble diminishes. Individual colony size (population) is linked directly to the velocity of the instrument associated with that colony. Birth/death rate is linked, in this case, to overall pitch set size, again ranging from one to nine octaves in 48 tone/octave equal temperament. Chaotic functions were chosen for pitch generation in an attempt to simulate some of the natural ambient sound found in the Colorado environment. The mechanisms for generating pitch sets, according to Lovell, are "nonlinear chaotic functions such as, xi+1=rxi(1-xi), that Mitchell Feigenbaum used when formulating his theories on chaos while working at Los Alamos National Laboratory in New Mexico [6]. This function is used in the interval where r is in the range from 0 to 1 and chaotic when r is > 0.86 and x starts as a low value. To generate a series of notes, an initial x value is fed into the equation and a new x is generated. This new x value is then scaled to match an audible MIDI note range and played. The new x value is then fed back into the equation generating a new note. The process continues at a certain metronome rate creating a melodic line. During the generation of new x values metronome speed and the r constant can be changed which forces the function to travel into new chaotic fields and structures.

example of a chaos function realized in MAX

In conjunction with the mural and the visual experience of the artificial world itself, the sound score completes the creation of a mediated reality, existing somewhere between the artificial reality inside the computer and the all encompassing world of our own natural environment.


Working at Deep Creek school provided the artists with the inspiration of a beautiful natural setting. The challenge of "bringing the rocky mountains into the computer" was an important element in the first performance of EIDEA. The concept of monitoring weather information stems from the artists' fascination with performance events based on real-time data input and an interest in exploring the rhythm of the natural world at Deep Creek. During the summer in the Rocky Mountains weather patterns exhibit a clear daily cycle, and one of the goals of EIDEA was to mediate and call attention to this ongoing natural phenomena. "This builds upon work done in previous years in which students and visiting artists have created works that respond to the local conditions--particularly Deep Creek itself" says Collins. "Mitchell and Lovell's discussions and interactive environment were enthusiastically received by the students and faculty. "

The artists plan to continue the development of EIDEA and to seek other opportunities for presenting the installation.

?[1] for more information see Mitchell, John D., (1990)Interactive Performance Works: the Computer at the Center of MIDI Based Performance Systems published proceedings, First Conference on Dance and Technology, Madison WI, pp. 41-45, 1990.

[2] L system description.

[3] Von Neumann, J. (1966) Theory of Self-Reproducing Automata, edited and completed by A. W. Burks, (Urbana, IL: U. Illinois Press).

[4] Reynolds, Craig W., "Flocks, Herds, and Schools: A Distributed Behavioral Model", Computer Graphics, V21, N4,
pp. 25-34, July 1987

[5] Alan Turing, "On computable Numbers..." reprinted in <Martin Davis, ed. The Undecidable, Raven Press, 1965.
[6] ÒPhysicist Mitchell Feigenbaum, predicted that at the critical point when an ordered system begins to breakdown into chaos, a consistent sequence of period-doubling transitions would be observed. This so called Ôperiod-doubling route to chaosÕ was thereafter observed experimentally by various investigators. Feigenbaum went on to calculate a numerical constant that governs the doubling process (FeigenbaumÕs number) and showed that his results were applicable to a wide range of chaotic systems. In fact, an infinite number of possible routes to chaos can be described, several of which are Ôuniversal,Õ or broadly applicable, in the sense of obeying proportionality laws that do not depend on details of the physical system.Ó

Gollub, Jerry and Solomon, Thomas. ÒChaos Theory,Ó The American Encyclopedia. Grolier Electronic Publishing, Inc. 1992

[7] Equipment
The Following is a complete equipment list for EIDEA
One SGI Indigo II XL with a 150mhz processor, 32 meg ram
with the Galileo video digitizing board
One Quadra 800 with and Expanse NUbus expander, Greenspring serial expander, two SampleCell cards, Lexicon Nuverb32 meg ram
Mackie 1604 mixer with Otto on board
Roland D110 sound module
Alesis Quadraverb
Two IVL Pitchriders
Protools 4 track recording system
MOTU MIDI Time Piece
8x8 videoswitcher, RS-232 controlled