ABB Robots and Home Robots That Work in Unseen Homes

From 1937 bed-making lessons to Figure's Helix 2.5 doing chores in 30 unseen homes, and what ABB's Item Picker and YuMi show about robots that adapt.

INDUSTRIAL ROBOTICS

Chat With Robot

9/26/20261 min read

Around 1937, home economics students in New South Wales practised making a bed properly, with tight corners and a smooth cover. Housework was learned by repetition in one house, and anyone who came to help had to learn that household's habits all over again.

Black and white photo of young women in white uniforms making a bed in a classroom bedroom
Black and white photo of young women in white uniforms making a bed in a classroom bedroom

A home economics class learns bed making in New South Wales, c.1937. Photo: State Records NSW (no known copyright restrictions)

Early home robots ran into the same wall. Research robots like Cosero at the University of Bonn could fetch drinks and water plants around 2012, but only in rooms they had been mapped and tuned for. Folding towels, done by hand in commercial laundries back in 1917, stayed out of reach.

Blue and white wheeled service robot holding a green watering can next to a potted plant
Blue and white wheeled service robot holding a green watering can next to a potted plant

Cosero, a household service robot from the University of Bonn, watering a plant. Photo: Sven Behnke / Wikimedia Commons (CC BY-SA 4.0)

In September, Figure reported that its Helix 2.5 model did chores in 30 Bay Area homes its robot had never seen, with no extra training. Over 420 trials of bed making, towel folding and toy tidying it succeeded 237 times, or 56%. Figure says pretraining lifted zero-shot success from 9%.

Sepia photo of a girl folding towels at a table beside tall stacks of linen
Sepia photo of a girl folding towels at a table beside tall stacks of linen

A 15-year-old worker folds towels by hand in a laundry, 1917. Photo: Lewis Hine / Library of Congress (public domain)

Factory robots face a smaller version of this problem. ABB's Robotic Item Picker uses AI vision so it can handle a changing mix of products in a warehouse without someone teaching every item one by one.

ABB robot arm with a suction cup gripper reaching down into a picking area
ABB robot arm with a suction cup gripper reaching down into a picking area

ABB's Robotic Item Picker with a suction gripper. Photo: ABB

When a job does need teaching, ABB tries to keep it short. The single-arm YuMi can be set up with the Wizard block-programming app on the FlexPendant, and a YuMi once made coffee for shoppers at Selfridges in London, working in front of the public.

Small white one-armed ABB robot beside a handheld teach pendant showing block programming
Small white one-armed ABB robot beside a handheld teach pendant showing block programming

A single-arm ABB YuMi next to a FlexPendant running the Wizard app. Photo: ABB

A 56% success rate is too low for a robot you pay for, though going from 9% to 56% in one step is a big jump. My guess is that the first home robots will arrive with a general model and then spend a few weeks learning each family's house, much like a new helper did in 1937.

Two-armed ABB YuMi robot preparing coffee at a counter in a department store
Two-armed ABB YuMi robot preparing coffee at a counter in a department store

An ABB YuMi making coffee for shoppers at Selfridges, London. Photo: ABB

Innovation

AI solutions for effortless ABB robot control.

Automation

Robotics

ceojohntran@chatwithrobot.net

+84905311611

© 2025. All rights reserved.

qtran1215@gmail.com

Instagram