I flew to China to visit a robot butler
Site visit to Prosper Robotics field lab in Shenzhen
Hello friends,
I’ve got something for you today which I’ve been working on for several months.
This summer I went out into the field, all the way to the other side of the globe in fact, to investigate a robotics startup that’s building robot butlers for your home.
For those of you who prefer to watch rather than read, I made a YouTube video about the experience.
The visit took three days. This write-up took two afternoons. The video, however, took me nearly four months to complete, and to be honest, all I can see at this point are the many ways I could still make it better.
It turns out that making YouTube videos is very difficult. I might do some writing about it in the future, maybe after crossing some click-friendly threshold like 10,000 subscribers. As of now, I have 42.
I’ve also published a duplicate version of this in the Bot Eat Brain newsletter, so don’t bother reading both if you’re already a subscriber over there.
Should I do more content like this? Hit reply and let me know what you think.
Onward to the juicy content…
DAY ONE
A robot that makes your bed.
The robots in this gif aren’t CGI, they’re an early prototype of a robot butler named Alfie.
These robots will cook for you, they’ll clean for you, they’ll even make your bed.
The team behind these bots is called Prosper.
I wanted to know:
1. What can these bots actually do?
2. How do they work?
3. and how much do they cost?
There’s only one problem: I’m based in the USA and they’re in Shenzhen. So, I flew to China to get my questions answered.
Wait, hold up, “Why do I need a robot maid in the first place? Can’t a human do the job just fine?”
Well, yeah. Humans are fast and effective at cleaning toilets and making food. These robots, at least for now, are slow. and frankly, still kinda clumsy.
But humans need to sleep. Humans are expensive. These robots can work 24/7 and, in theory, work for a lot cheaper than humans.
Cheap enough that someone who could never afford a human maid COULD afford a robot butler.
But cleaning up after your messes is just step one for these bots.
Because if you can build a robot capable of keeping a house in order, then you can build a bot that can do just about… anything.
Things like: construction, handling hazardous waste, diffusing bombs, or any other job you can’t find enough humans for or that we’d rather not do in the first place.
And while the current version might be slow today, Alfie is still a work in progress.
This robot is under construction and the team is racing to get it ready for their first customers sometime next year.
The robot will stand about 170cm tall and weigh more than 100kg. That’s around 5ft 6in and 200 plus pounds.
The bot consists of a base on wheels with two mechanical arms and pinchers for hands.
Legs are considered unnecessarily complicated. Stairs can be conquered with a lift system or (if ya rich) you could simply buy a second bot.
Ditching the legs should save years of development time.
At first, the bot will be almost entirely controlled by a person, [UPDATE: the robot will now be partially autonomous from launch], who will basically be playing a video game in VR, where the game is to clean your house via remote control.
As long as they have fast internet, they can work from anywhere.
Prosper will gather data from their early bots to make their AI work better. Over time, the robots will be driven less and less by humans and more and more by code.
For now, human supervision will keep the bot from pulling an iRobot and (much more likely) will help handle mundane situations like figuring out a new dishwasher or finding your sheets.
The team says that you can expect a human to be in the loop for a while still.
First comes the platform: a robot that is physically capable of doing what you need.
Only then comes the AI: that is, the software that can drive the robot for you.
In the beginning, humans will drive the robot completely to make sure it does its job safely and well.
As the software improves, humans will need to directly control the bot less and less.
Eventually, human drivers will only need to be available to help the robot get “unstuck” when there’s a task it doesn’t know how to handle.
DAY TWO
Visiting the world’s largest manufacturing hub.
The team is in China right now to find the materials and suppliers they need to manufacture that first small batch of bots.
I spent my first day in Shenzhen shadowing the Prosper team in their small field lab full of humming 3D printers and jumbled boxes of test parts as they worked on the supply chain of their robotic butler.
By coming to Shenzhen instead of hanging out at their London HQ they’re able to get samples and quotes much more quickly than they could back home.
What would take weeks, gets cut down to days. Now multiply that times hundreds of individual parts and you start to understand why they’re willing to spend months flying back and forth from halfway across the world.
On my second day in China, I followed the team on factory tours around Shenzhen.
Our first stop was at a small shop that seemed like they could make a little bit of anything. They had dozens of different methods and materials to choose from:
injection molding, CNC routers, you name it, they got it.
Our tour guide was very agreeable but there was always the feeling that we might just be getting the answers we wanted and not what the team really needed to know.
That’s why it’s so useful to be there in person and see things for yourself.
The second stop was a metal stamping factory, it was pumping out hundreds of small metal pieces every minute. Way cooler on camera. Much less relevant for prototyping robots.
So, when can I buy one?
Not so fast.
An earlier version of the bot is already being tested in the homes of Prosper employees, but the first robots won’t be available to buy until sometime next year, and even then it’s only going to be a few people that get to buy one at first.
You’ve gotta be in London, no kids, no pets, no stairs.
Oh yeah, and the starting price? At least $5,000 (and that’s not including the $500 monthly subscription).
The cost will cover the robot, its maintenance, and the salary of the human drivers.
So Version 1, is definitely not meant for everyone.
These beta testers need to be close enough to the main HQ in London for frequent robot check-ups and be willing to put up with some robot nonsense as the team works out the kinks.
As soon as the robot is done, you better believe I’m flying to London to check it out for myself, subscribe to my YouTube channel and you’ll be the first to see it.