Results from the second-annual Robot Art Competition, a celebration of beautiful artwork created by A.I., are finally in. The winner and recipient of a $40,000 grand prize was Columbia University, with their vivid, robot-created oil paintings of a man’s face, a distant house, and a dancing flame.
The contest was created by Andrew Conru, an engineer who saw painting and artwork as a way to better integrate robotics into society and culture, and this year’s contest ranked submissions from 38 teams who represented ten different countries. Conru hopes that the contest will foster new developments into machine learning.
There were no set guidelines for how the robots had to create their masterpieces, and the teams tended to take a variety of different approaches. Some employed motion-capture systems where a robotic arm mimicked a human painter or used manual controls to operate the robot, with the person behind the controls painting by robotic proxy.
But other teams tried some more creative techniques, such as using eye-tracking software to guide their robot or by training their A.I. to analyze a photo or image then try and recreate it while mimicking different art styles. Yet other A.I. painters leaped into the realm of the abstract and turned algorithms into brushstrokes to create totally new images, such as Halmstad University which read a participant’s brain waves and converted them into art.
But without further ado, here are the top ten finishers for the 2017 Robot Art Competition and their award-winning artwork.