AI art is currently on the rise. Although there is still a heated debate concerning art's originality and integrity, several AI tools are introduced to the public now and then, promising innovation. We cannot help but be stunned by these cutting-edge technologies.
And this time, a robot is trying to replicate the magic of AI art generators. Carnegie Mellon University (CMU) Robotics Institute does this with FRIDA (Framework and Robotics Initiative for Developing Arts).
(Photo : @FridaRobot/ Twitter)
Some of FRIDA's (Framework and Robotics Initiative for Developing Arts) paintings.
FRIDA the Robot Painter
The name "FRIDA' is an obvious reference to one of the finest and most popular painters from Mexico, Frida Kahlo.
CMU professor Jim McCann told TechCrunch that this robot painter combines human and robotic creativity. It is equipped with models similar to generative AI tools that produce artwork from specific text prompts.
And as such, FRIDA requires input before creating its magic. It produces artwork based on text descriptions and even existing photos.
It is also worth noting that humans are not excluded from FRIDA's artmaking process. They are still involved in not just suggesting text inputs but also paint mixing and providing materials.
The robot's developers believe that it could serve as the helping hand of an artist. You can check out FRIDA's paintings through its Twitter page.
Some of Frida's first paintings! More to come pic.twitter.com/4ZsW95GnFn
— FRIDA Robot Painter (@FridaRobot) August 24, 2022
Simulation Environment for Painting
According to the team's paper published online at arXiv, they developed a completely different simulation environment for painting. They also adopted the concept of reality to simulation to real (real2sim2real).
The developers claim that their proposed simulated painting environment is closer to reality than other simulation environments employed for robot painting.
The team also crafted a planning strategy that can continuously improve the painting plan based on the changing canvas to reflect the changing dynamics of a creative process.
FRIDA has a dynamic nature in creating artwork by constantly re-planning and re-assessing its semantic objectives via its visual perception when painting. In short, the robotic painter's actions are not pre-determined; they can be spontaneous.
This is in stark contrast to current approaches where content generation is carried out independently and sequentially, according to the robot's developers.
FRIDA is one of the existing tools, riding the generative AI trend. This general label is used for any AI software that produces new images, texts, videos, and other content.
But the difference with FRIDA is that content generation is limited to the physical robot, unlike tools like StableDiffusion and Midjourney, which are both accessible online.
Analysts estimate that the generative AI market will boom even further and become a massive $110 billion market by 2030.
Related Article: Real Estate Brokers Now Use ChatGPT for Emails and Property Listings; Experts Claim AI Could Become an Industry Standard
