The Aeolus robot is supposed to be everything that overtired parents have dreamed of. Using mechanical arms and machine learning, the robot can pick up stray toys and put them into bins, clean the floors with a standard vacuum, and even help owners unwind by grabbing a beer from the fridge.
“By my estimation, in the future, every family will have at least one robot,” says Alexander Huang, the founder of Aeolus Robotics.
Aeolus wasn’t the only company to promote that vision at the CES trade show this month. Other robots on display promised to fold your laundry, look after elderly loved ones, or keep a lookout for intruders. Although robots have long been a CES fixture, they’re often for education or industrial use, or are just variations on the Roomba. This year, as artificial intelligence becomes more advanced, the robots edged ever-closer to the Jetsons-inspired vision of a machine that does all your gruntwork.
Even so, the robots on display at CES were either experimental in nature, prohibitively expensive, or both. For average consumers, capable smart home robots are at least another five to 10 years away.
Learning To Work
Huang, who was previously the head of Microsoft’s China unit and the Chief Strategy Officer at component maker Lite-On, says Aeolus has only become possible in recent years because of cheaper components and advances in artificial intelligence. The company was founded in late 2016, and plans to ship its first robot–still to be named–by the end of this year.
Still, the robot remains a work in progress. At CES, Aeolus sometimes had trouble figuring out what to do, prompting the demonstrator to pick up and reposition objects in hopes of triggering the robot’s recognition. Even when Aeolus was working as advertised, its arms and wheels moved at a glacial pace. Picking up a stuffed animal and placing it in a nearby bin took about one minute, and this was in a contained environment, on a clean wooden floor with only one other object in the vicinity.
Huang says the robot can technically move faster, and that the current speed is a safety measure. But he also acknowledged that the robot’s AI-based software needs more training, which is the main thing Aeolus intends to work on before the robot launches. “When this robot learns, the rest of the [Aeolus] robots also learn immediately,” Huang says.
Even if Aeolus can smooth out its robot’s rough edges, the product won’t come cheap. Aeolus says it’ll cost about as much as an overseas family vacation–the company won’t give a specific price just yet–and users will also have to pay extra if they want more than a year of maintenance for any problems that might arise. Huang expects that customers will replace their robots outright after just three to five years. Asked when this kind of robot might become more affordable, Huang instead extolled the benefits of spending big on a robot who can help with chores.
“I would say that time is really valuable,” he says.
“To liberate people from laundry work”
Aeolus wasn’t the only robotics company at CES promising to handle household duties. Foldimate and Seven Dreamers were both tackling folding clean laundry, with machines that could fold clothes on their own.
Foldimate, the cheaper of the two, with an expected price of around $1,000, comes with a catch: Users will have to feed each garment into the machine one-by-one. That could be nearly as time-consuming as actually folding the clothes themselves. Seven Dreamers’ Laundroid was more sophisticated, allowing consumers to drop an entire pile of laundry into the machine, but that luxury will cost about $16,500 when it arrives in the United States next year. (Laundroid has already launched in Japan for well-heeled consumers at the same price.)
Shin Sakane, Seven Dreamers’ president and CEO, says picking garments out of a pile is a major challenge for Laundroid’s robotic arms. To recognize clothing with 95 percent accuracy, Seven Dreamers had to take 256,000 photos for each type of garment. The system then has to spend time pinching and pull at each article of clothing, figuring out what garment it’s holding and how to fold it.
“If you put a piece of clothing in a certain place to fold, it’s easy,” Sakane says. “But from there, it’s so hard.”
Seven Dreamers’ goal is to bring the price under $2,000 within five years, both by manufacturing at a larger scale and adding services to help defray the hardware cost. In Japan, for instance, Seven Dreamers has partnered with fashion startup Aircloset to provide recommendations on what to wear, and Sakane said the company is looking into similar partnerships in the United States.
“Our vision is really to liberate people from laundry work, and if it was always $16,000, only limited people can participate,” Sakane says.
Still, Seven Dreamers doesn’t any track record making consumer products at scale. Beyond Laundroid, the company’s only other products are luxury carbon golf club shafts and a nasal cavity insertion device for sleep apnea sufferers. Sakane hopes Seven Dreamers will someday become a publicly-traded company with the cachet of an Apple or Sony, but it’s got a long road ahead to get there.
Compared to the flashy demos from Aeolus and Laundroid, Misty Robotics’ personal robot is less impressive on the surface. It’s only about 15 inches tall, has no mechanical arms, and doesn’t do much out of the box. Yet Misty, which was spun off from BB-8 toy droid maker Sphero, has devised a more methodical approach to building smart home robots. Instead of starting with an expensive consumer product, Misty is pitching its first robot at developers, who can pay $1,500 to start building their own “skills.”
“We’re taking our queue from every historical large technology disruption, like the web or personal computers, where they started with programmers,” says Tim Enwall, Misty’s CEO.
Enwall figures there are plenty of potential uses for a robot that only includes a camera, microphones, a speaker, a display that makes faces, and wheels for getting around. Developers might create a skill that looks after an elderly relative at home, for instance, or one that checks supply room inventory at the office. But Misty will also include serial and USB ports, which would eventually allow for robotic arms or other extremities.
The problem, says Misty co-founder and head of product Ian Bernstein, is that robot manipulation just isn’t practical for consumers today, requiring high-precision motors and expensive bearings. He’s hoping that over time, more computing power will allow for motors that are just as accurate at lower prices, even if they’re a bit less stable. By starting with the basics–eyes, ears, mouth, and mobility–Misty can at least begin creating a platform.
“We’re really confident that that’s a rich enough invention space for the next few years, while the technology and cost of manipulation comes into a consumer-affordable price point,” Enwall says.
That doesn’t mean Misty will fulfill your Jetsons robot dreams any faster than other companies. Enwall says the company is working off a 10-year plan, the first few years of which will cater entirely to developers. Early adopters might get interested around year four, but they may still have to wait a few more years for mechanical arms and other appendages to become practical.
In the meantime, Boulder-based Misty has amassed $11.5 in venture capital investments led by Venrock and Foundry Group, and Enwall says they’re on board with the long-term plan.
“They know that consumers aren’t interested or ready for robots,” he says. “They know how the web was developed. They know how personal computers were developed. They know this is a 10-year or longer arc.”
So, sorry parents: you’ll be folding your own laundry and picking up stray toys for the foreseeable future. But hey—at least robots that can play your music or vacuum your floors have never been cheaper.