Inventor James Dyson has a long history of re-imagining the familiar. From bag-less vacuum cleaners to touch-free hand-dryers in bathrooms, these are devices that do something commonplace, but in a way no one thought of before or imagined could work. Now, with Dyson’s investing more than $8 million in a joint robotics lab at Imperial College London, some might think Dyson is branching out. They’d be wrong.
First of all, Dyson started working on robotic technology 16 years ago and it was primarily devoted to the development of — that’s right — a robot vacuum.
See also: Dyson's DC59 Vacuum Runs 5 Times Faster Than a Race Car
“We actually tooled up for a robot,” Dyson told me. The company built, demonstrated and showed off a robot vacuum way back in 2001 (a year before iRobot’s Roomba robot vacuum became a consumer sensation). It was the DC06 and featured more than 80 sensors and two on-board computers, and if it had made it to market it would have cost $3,000.

“We decided not to launch it partly because of the price and also, like all engineers, we had a better idea,” said Dyson.

Over the next five years, The Dyson Robotics Laboratory, led by Lab Director Professor Andrew Davison, will work on developing vision-based robotics systems that could lead to a whole new generation of self-aware, human-friendly robots.
However, near-term, it will probably lead to a new vision-based Dyson robot vacuum. This would make perfect sense, since, even though Dyson abandoned his first robot project more than a decade ago, he never really abandoned the vision.
While Dyson was developing his first robot vacuum, he realized that they these intelligent appliances faced a significant hurdle. They had to clean perfectly while managing a typically limited resource: battery power. “You want to clean properly and not go over the same spot twice, you can’t afford to,” said Dyson.
Dyson became obsessed with perfect navigation and not wasting any energy. The company, however, realized the best navigation might come from a vision-based system instead of a sensor-based one and, Dyson told me, they realized battery technology would eventually get better.
In the end, the company switched gears, as it were, and turned its attention to refining its electric motors. The company now builds millions of these so-called "digital" motors in an automated, robotized production line. They currently appear in Dyson’s Digital Slim Vacuum.

Sometime in the not-to-distant future, however, Dyson will launch a robot vacuum that relies on this new motor and, more importantly, a vision system that may deliver “perfect” navigation.
“Vision interpretation is the key to the future of robotics,” said Dyson.
Dyson insists that robot vacuums that use a combination of room mapping and obstacle detection (usually through touch sensors, but also in tandem with virtual wall beacons) offer “very, very poor dirt removal.” “You might cover the majority of floor, but in the process you haven't been able to apply any decent amount of suction or pickup,” added Dyson.
No one wants to vacuum, but few would argue that robots have yet to achieve human-like efficiency at getting the job done. iRobot’s arguably effective Roomba usually takes longer to complete a room than a person vacuuming the floor with, for instance, one of Dyson’s own bag-less vacuum cleaners. On the other hand, the idea is that you don’t have to stand around and watch your Roomba vacuum the floor; you simply set it and let it vacuum on a schedule — usually while you’re not there.
A robot with the ability to “see” a room and identify all the objects in it, however, might do a better job, at least according to Dyson.
Dyson and Imperial College have, in fact, been working on such a system for years though now, with the added resources, they’ll add dozens of scientist and researchers to help accelerate work on the vision system.
That system relies, in part, on SLAM, or Simultaneous Localization and Mapping (one of the technologies commonly used by self-driving cars), combined with a library of information about objects. That combination will help Dyson robots act to a certain extent more like humans. “We humans don't see all of an object, but we know that object because we have the history of it,” Dyson said. So when you look at the corner of a couch, you brain knows that there’s seating on the other side. That information will affect how you interact with the couch.
Dyson wouldn’t reveal what other kinds of robots they’re working on beyond a robot vacuum, but did offer, “If we are successful developing vision systems, the scenario for applications open up.”
Growing up in the 1950’s, Dyson had no trouble imagining robots replacing humans across a wide range of tasks. That sci-fi future, though, has been painfully slow in coming. “The fact of the matter is, you try to start to make [a robot] and you see how complex it really is.”
Even today as Dyson and his partner researchers prepare to bring vision-based robots to the masses, limitations and hurdles remain.
For one thing, these robots are not truly autonomous. You still need, for instance, to clean out the robot vacuum. “A robot that cleans itself? Maybe one day. For now you need another one that sucks [all the dirt out] and then you need another one to clean that one,” Dyson laughed.
He also told me that our robot future may require that we change our homes and offices to make them more robot-friendly.
And then there’s the price. Vision-capable sensors can be expensive. Dyson knows what people are currently paying for robot vacuums and recalls his company’s own history with the $3,000 DC06. How will he price the next Dyson robot vacuum? “I don’t think [a robot vacuum] is for everyone at all,” he said. But there are those “who are tickled by the idea of a robot doing something” instead of them. “They’re willing to pay quite a bit for it. The more intelligently and expertly it does [its work], the more they’re willing to pay.”

Have something to add to this story? Share it in the comments.