Robot artist uses AI to paint in the style of great masters

The robot is named Frida, after Frida Kahlo, which also stands for Framework and Robotics Initiative for Developing Arts - Carnegie Mellon University

The robot is named Frida, after Frida Kahlo, which also stands for Framework and Robotics Initiative for Developing Arts – Carnegie Mellon University

Possessing little artistic ability may no longer be a barrier to creating great art, after scientists designed a robot that can paint their ideas into reality.

Carnegie Mellon University has developed an artificially intelligent robotic arm that will paint on command once given an idea.

Similar to how ChatGPT generates text in a chosen style, users can direct the bot to paint a specific image, copy a photograph, or work in another artist’s style.

The robot is named Frida, after Frida Kahlo, which also stands for Framework and Robotics Initiative for Developing Arts.

“Frida is a robotic painting system, but Frida is not an artist,” said doctoral student Peter Schaldenbrand of Carnegie Mellon’s Robotics Institute, who helped develop the robot.

“It is not generating ideas to communicate. It’s a system that an artist could collaborate with. The artist can specify high-level goals for Frida and then execute them.

He added: “There is a painting of a dancing frog that I think turned out really well. It’s really silly and fun, and I think the surprise of what it generated based on my input was really fun to watch.”

Frida has been trained on massive data sets that combine text and images pulled from the internet, before AI systems process the data to generate a new image that she can then paint.

Once Frida has been given a concept to work with, the robot uses machine learning to create a simulation of the final painting, then develops a plan for how to achieve it using brushstrokes.

It also uses machine learning to assess your progress as it works. From time to time, the robot uses an overhead camera to capture an image of the painting, refining its plan if necessary.

The team describes the final paintings as ‘impressionistic’ and ‘whimsical’ which curiously lack the precision expected of robots.

The researchers say they are struggling to improve Frida’s artistic ability and admit there is room for improvement in what she composes in the simulation compared to the final result on the canvas.

Even if the program improves, though, the team doesn’t think artists are in danger of being replaced by AI-powered robots.

“People wonder if Frida is going to take jobs away from artists, but the main goal of the Frida project is the complete opposite. We want to really promote human creativity through Frida,” said Dr. Jean Oh, a research professor at the Robotics Institute.

“For example, I personally wanted to be an artist. Now, I can collaborate with Frida to express my ideas in painting”.

Dr. James McCann, Assistant Professor at the Robotics Institute added: “Frida is a project that explores the intersection of human and robotic creativity.

“Frida is using the kind of AI models that have been developed to do things like caption images and understand the content of the scene and apply it to this artistic generative problem.”

Leave a Reply

Your email address will not be published. Required fields are marked *