BBCThe idea of having a friendly robot butler that can do all the dull duties of running a home has existed for decades.
But now, thanks to AI, it’s genuinely happening and this year the first truly multi-purpose domestic bots will start to enter homes.
In Silicon Valley, they’re being trained at speed to fold laundry, load the dishwasher, and clean up after us.
Their excitable human creators are making big promises but I wanted to see how realistic the idea of a robot housekeeper really is.
So I went to meet Eggie, NEO, Isaac and Memo.
Tangible AIIt is impossible not to smile when one of these humanoid or partly humanoid (no legs) bots enters a room.
The overall state of play is that many of them are now agile, sensitive and dextrous enough to carry out many important (and tedious) chores.
We watched as Eggie the robot from relatively fresh start-up Tangible AI hung up a jacket on a coat stand, stripped a bed and wiped up a spill on the kitchen counter.
But it did it very slowly, rolling around on wheels in a stuttering movement.
Likewise NEO from 1X – which recently caused a stir by launching pre-orders for its robot – was able to slowly but effectively plod around the firm’s test kitchen on its soft padded feet.
It watered plants (with one spillage), fetched me a drink and tidied away dishes and cups (with some help from me as it struggled to grip the cupboard handles).
If time was no issue, I could see how having an Eggie or NEO-like bot cleaning up after me and my kids might be helpful.
But NEO and Eggie have a secret weapon – they are being controlled by human operators.
This is the thing the promotional videos don’t show – and something that the Silicon Valley companies we visited are keen to downplay.
Bipasha Sen, founder of Tangible AI, is upbeat though about how fast the tech is improving.
“Today people have two aspirations – a car and a house. In the future they’ll have three aspirations – a car and house and a robot,” she says with a beaming smile.
Across town, 1X is a company that has major financial backing from tech giants including microchip maker Nvidia.
1XAt their plush headquarters, we were given a tour of a restricted area where NEO prototypes are being built, tested and repaired.
Norwegian CEO Bernt Børnich says NEO is very useful in his own home, busily hoovering and tidying up after his family, which he says is “a mix” of autonomous action and human-operated.
“We have a lot of data so a lot of the stuff in my home can get automated but periodically someone kind of steps in and helps,” he says.
Data is key to how these robots are learning to navigate our chaotic home environments – a much tougher task than humanoids designed for factories.
Part of 1X’s plans to improve NEO’s AI brains is to get it out to homes this year.
1X is confident that NEO will be far more capable on its own thanks to recent AI developments.
But we weren’t shown any demos of the bot thinking for itself.
The first wave of customers will probably have to be very patient and not that worried about privacy with human operators remotely controlling it when the bot gets confused.
They will also have to be wealthy as NEO will cost around $20,000 or $500 a month.
“A lot of our early customers are people who will actually have a lot of value from this, but I do think getting the right customers is important. We can use these amazing early adopters to help us make this work,” Børnich says.
Unusually for tech, most investment and hype around household robots seems to be going to start-ups – not the tech giants.
Tesla is building a humanoid but it is not clear what market it will be aimed at – factories or homes.
CEO Elon Musk is convinced there will be a big market for them though – his record $1 trillion pay packet is partly linked to him selling one million bots in the next ten years.
But it’s nimble Silicon Valley start-ups that seem to be best placed to hit the market first.
In Noe Valley in San Francisco, another domestic robot company has already deployed its stationary bot to gather real world data, albeit in the narrow task of folding laundry.
Weave Robotics has seven Isaacs dotted across the city, autonomously folding clothes for laundromats.
We watched it meticulously fold T-shirts in about 90 secs, but its creator says it is getting faster all the time.
“Deployment is the strategy,” says co-founder Evan Wineland.
The company plans to launch a general purpose version of Isaac for homes this year, but it’s not clear how many tasks will be autonomous.
Elsewhere at Sunday AI they’ve come up with a neat solution to the data collection challenge that seems to be working very well.
We watched its robot slowly but smoothly make a coffee, scrunch up some socks and clear a table of perilously fragile wine glasses. All on its own with no human intervention.
But even this highly capable bot made one mistake – breaking a wine glass on its first attempt, which appears to have been a bad fluke.
Engineers here are confident all will be ironed out once the bots ship next year thanks to a robot glove they’ve developed.
“We built these gloves and people wear them in their homes and collect data for us and that gives us really diverse data because we now see 500 homes and also all the different ways people do chores,” says co-founder Tony Zhao.
It’s a reminder of the human drudgery underpinning how AI systems operating in the physical world learn.
Teaching AI chat bots is easy in comparison as they are able to absorb billions of web pages, books and films to get smarter.
The last company we visited has a completely different angle on how to make the domestic robot a reality.
Physical Intelligence isn’t interested in making a robot itself – it’s developing the brains to make dumb robots smart.
Engineers are using all sorts of different robotic arms, hands and bodies to develop AI software for any robot hardware.
Physical Intelligence“We want to be able to breathe intelligence into any sort of physical embodiment, whether that’s a humanoid robot or even something that looks closer to an appliance,” says co-founder Chelsea Finn.
Their approach is being excitedly backed by investors including Amazon founder Jeff Bezos and OpenAI.
There’s a huge amount of investment going into this technology and although Silicon Valley is once again an epicentre, it is facing tough competition from Chinese rivals.
The industry for humanoid robots in general in China is in fact so hot that the government recently warned there was a risk of a bubble building that might burst if the robots aren’t as successful or popular as they hope.
The International Federation of Robotics thinks it could take 20 years before domestic bots become truly useful and accepted.
There are questions too about how much demand there will actually be for the bots. Will they just be the play things of the rich or will they become cheap enough for mainstream use in the same way that robot hoovers have become?
But for the engineers at the forefront of this technology there appears to be a confidence that they are truly building a future that all of us will want in our homes.