The combination of art, technology, engineering and artificial intelligence has had amazing results, allowing artists of painting, sculpture, dance, theater, among others, to have a tool that allows them to advance towards their creative objective with greater efficiency, precision and speed.
Two researchers from the Vienna University of Technology Eric Steinberger (self-taught AI research expert researches in the fields of artificial intelligence and robotics) and Benjamin Mörzinger (TU Vienna researcher, their commitment is to accelerate the transition to Industry 4.0 with innovative methods). The goal of this research is to paint pictures using various techniques and to generate the concept to order paint and water to clean your brushes.
This duo managed to demonstrate the unlimited capabilities of an industrial robot, with massive customization they achieved flexibility using AI and dynamically generated motion series in the simulation, generating new opportunities.
“Smart” as this creative painting project was called includes an ABB IRB 120 painting robot combined with artificial intelligence and using genetic algorithms, the robot is taught to paint with acrylic paint.
Smart processes the photograph by means of a computer. The computer stylizes this photo in a graphic style or the style of Van Gogh or Picasso. Image content is digitally recreated in one of these styles. A very simple genetic algorithm with manual selection and no crossover is used to quickly create and test strokes without having to manually design them, this way the robot learns to use brushes to create visually pleasing lines. One of the robot’s tasks is to fill paint and also order paint. additionally, you have to drink water to clean your brush. The result was a visually attractive image of a male face that was painted by the robot without any human intervention with acrylic paint and brushes. The learning process had special security conditions, since the robot is experimenting. The process of learning brush strokes is very efficient in time.
The technique that was used consists of two systems. One system is the robot, the other is its collected periphery, which we call the paint factory. It gives you the ability to interact with your environment, exchange data and cooperate.
This system incorporated components such as creating a library of brush strokes interactively.
The robot’s own style is applied to each image it will paint via Neural Style Transfer (NST).
The combination of strokes to achieve an image style through simulation. Using Iterative Trace Sampling (ISS), the ISS implementation is to get a list of commands for the ABB robot to execute. Via the ABB RAPID Code Interface (ABB Robot Programming Language). Where automatically the RAPID code is saved to disk with the name of the initial image.
In an interview with Eric Steinberger he said, “The goal was to show that robots can not only carry out monotonous orders and sequences of movements, but also creative ones.”
The robot is still a tool that executes a trajectory but in combination with AI and software and other intelligent functions, it is suitable for various tasks in different sectors to the point of being an assistant or useful tool for an artist.