Tired of laundry folding? AI breaks the robot folding speed record

It's likely that someone out there doesn't enjoy folding clothes. SpeedFolding is a new method of folding garments for a robot.

SpeedFolding can fold 30–40 randomly positioned garments per hour, finishing each one within two minutes.

While that rate does not sound impressive compared to a human, previous robotic garment-folding methods reached only "3-6 FPH", according to the researchers in a paper submitted for presentation at IROS2022 next week in Kyoto. It's the work of several people, including Ken Goldberg of UC Berkeley.

The paper describes the garment-folding problem in technical terms and is worth a read. They write that "Folding garments reliably and efficiently is a long-standing challenge in robotic manipulation." Clothes can be all over the place because they are easy towrinkle up. They came up with a manual system that could smooth out a crumpled garment and fold it along preferred folding lines.

The authors say that previous robotic garment-folding methods have mainly focused on single-arm manipulation.

After studying 4,300 human and machine-assisted examples, the BiMA-Net neural network was able to fold clothes from a random initial position in under 120 seconds. SpeedFolding can apply what it has learned to clothes of a different type than the clothes it trained with.

The system needs to examine the initial state of the garment and calculate where to grab it with its two arms to get it to the next step in the folding process.

The researchers used a robot with 3D printed teeth to improve grasping. Dreams of running SpeedFolding at home are not currently within the reach of most people. The technique could eventually be used to make householdrobots.

Potential applications for the technology includetextile manufacturing and logistics, industrial and household laundry, healthcare, and hospitality. This aspect of our potential future can't get here soon enough for people who don't like folding and sorting clothes.