AI-enhanced robot from Goldberg’s Automation Lab can collect laundry

AI-powered robotic arm on a platform holds three pieces of clothing in the air with a white laundry basket on the right. The demonstration is done against a black background.
Photo courtesy of Ken Goldberg et al.

Out of the laboratory of UC Berkeley professor Ken Goldberg, director of the CITRIS People and Robots legacy initiative: Researchers have come a couple steps closer to building a robot that can help clean your room.

Combining cameras and clothes-recognizing AI, Goldberg and his team have developed a robot that can efficiently pick up strewn clothes. The integration of AI into their latest model allows the robot to perform tidying movements that minimize the number of trips it makes to the laundry basket, making it nearly 70 percent more efficient than preceding models that employ random collection.

The technology does not yet address further tasks such as sorting the clothes, and researchers are perfecting more nuanced robotic actions such as grasping multiple items simultaneously and using the proper amount of force to hold them. Even though these machines grapple with tasks that seem straightforward to humans, each step is progress towards making a robotic cleaning system household ready, says Goldberg.

Goldberg and his team previously developed SpeedFolding, a laundry-folding technique which uses two robotic hands to reach 30 to 40 folds per hour.