© 2024 Innovation Trail

Robots Are Really Bad At Folding Towels

Seven years ago, Pieter Abbeel set out on a quest: to teach a robot how to fold laundry. This proved to be a remarkably difficult task — and the difficulty of the task illuminates some key things about the limits of machines.

Abbeel, a professor at the University of California, Berkeley, named his robot BRETT — short for the "Berkeley Robot for the Elimination of Tedious Tasks."

For a robot, it's remarkably hard to figure out what's going on in a pile of laundry — to see, say, where the underwear stops and where the towel begins. Every pile of laundry is different, and remarkably complex.

Abbeel's team spent months staring at laundry baskets, holding towels up in the air, and taking pictures of laundry.

The solution was super complicated. "Can you use multiple images to build a 3-D model of the current shape?" Abbeel says. "Because once you can do that, then you can analyze that 3-D shape [and] find where the corners are."

Abbeel and his colleagues solved the problem, sort of. After years of work they taught BRETT to fold a towel in 20 minutes — eventually he learned to do it in a minute and a half. But he can still get stumped by things like a bundled-up sock or an inside-out onesie.

In other words, years of work from dedicated, smart researchers have produced a towel-folding robot that can't keep up with an average 8-year-old. This problem, Abbeel says, is not limited to towels.

"Once you start working in robotics," he says, "you realize that things that kids learn to do up to age 10 ... are actually the hardest things to get a robot to do."

Lots of things that seem simple and orderly to us are incredibly complex and chaotic for a robot. Machines need clear rules. One of the ways to figure out if a robot is going to take your job is to ask yourself: What are the rules here? Is my job a series decisions based on an orderly pattern? Or is my job really more like a giant pile of messy laundry?

Copyright 2023 NPR. To see more, visit https://www.npr.org.

Tags
Steve Henn
Steve Henn is NPR's technology correspondent based in Menlo Park, California, who is currently on assignment with Planet Money. An award winning journalist, he now covers the intersection of technology and modern life - exploring how digital innovations are changing the way we interact with people we love, the institutions we depend on and the world around us. In 2012 he came frighteningly close to crashing one of the first Tesla sedans ever made. He has taken a ride in a self-driving car, and flown a drone around Stanford's campus with a legal expert on privacy and robotics.