A Milestone for the Fusion of Art and Science
(24” x 42”) 3500 brushstrokes, 2003
A True Milestone in the Fusioneering of Art and Science and Robotic Art
The year was 2003. Dulcinea’s first painting, after she became fully automated, represents a milestone in the evolution of art and robotics. She became a full collaborator, not just a fancy printer.
With these added features, Dulcinea could truly paint unattended.
1) reach for and select from dozens of colors
2) wash the brush between colors, and to prevent the brush from drying out
3) reseal each color after its use, preventing each paint from drying out.
She started this painting one day in July 2003, continued painting during the night and finished in the early morning hours of the second day. There were 3,500 brush strokes in the painting. The color changes and brush washings were entirely automated.
A Surprise Ending That next morning, Paul Kirby unlocked the door and guardedly entered his studio to see the results. Dulcinea had indeed completed the painting, and had returned the last paint color, resealed the paint, washed the brush one last time, and returned to her “home” in the lower-left corner. Upon completion: she had also turned off her own power! This last-minute coding addition was astounding to behold that early morning.
AI Generated Painting Design This robotic painting used these forms of artificial intelligence to create the art:
1) genetic programming, a form of artificial life,
2) complex adaptive system simulations,
3) rule-based, backward-chaining expert system,
4) and some fuzzy logic in the rules set.
AI Genetic Programming, a Form of Artificial Life The painting’s design was created using genetic programming, a subset of evolutionary computing and artificial life. In genetic programming, the solution source-code program actually evolves through a series of “Darwinian evolutionary survival of the fittest” steps. The final outcome may even provide powerful and surprising results. (Ref “Genetic Programming” by John Koza.)
Painting’s Background: Initially, two colonies of ants were selectively bred and roamed around the field of the painting on the computer, creating the background patterns for the painting, including obstacles and ridges.
Painting’s Foreground: Then three other small colonies of ants evolved and created the brown and orange foreground, or main patterns of design. These sets of patterns were then converted into brush strokes.
AI Complex Adaptive System While genetic programming created the ant species, the simulation of wondering ants, using micro rules of behavior, formed a complex adaptive system. A collection of simple units or agents (our ants) interacting with each other leads to an overall emergent outcome. The fascinating aspect of this is the outcome is often difficult or impossible to predict by simply observing the individual interactions. Happy surprises are in the making!
AI Creates Color Palette Paul generated the color palette utilizing a rule-based, expert system with some added fuzzy logic.
The AI program queried the user (Paul) for what type of palette he wanted:
1) mood (happiness was selected)
2) style (Flemish was selected)
3) key (high was selected)
Each time the expert system was run satisfying these parameters, it created a unique color palette. The addition of the fuzzy logic ensured a subtle, varied, and fascinating outcome.
AI Summary Unique art was created by
1) creating ant species within genetic programming,
2) applying them within a complex adaptive system of micro rules,
3) and creating an individualized, expert-system color palette.
Conclusion – A Milestone in the Fusion of Art and Science
Thus, in July, 2003, this painting marks a true milestone in the evolution of art and science and robotic art.
It was fully autonomous in its painting and various forms of artificial intelligence created its unique design and color palette.