A robotic system for interpreting images into painted artwork

We report on a robotic system that can physically produce paintings with a wide range of artistic media such as acrylic paint on canvas. The system is composed of an articulated painting arm and a machine-learning algorithm aimed at determining a series of brushstrokes that will transfer a given electronic image onto canvas. An artist controlling the system is able to influence the resulting art piece through choice of various parameters, such as the palette, brush types and brushstroke parameters. Alternatively, an artist is able to influence the outcome through varying the algorithmic parameters and feedback of the learning algorithm itself. In these results, a genetic algorithm used a painting simulation to optimize similarity between the target and the source images.