Robots have not established themselves as helpers in everyday life – apart from vacuum cleaners. Why is that and what has to change for that to change.
They’re bland, quiet, and at the pushiest when they’re wrestling with the edge of a carpet or getting tangled in power cords. Vacuum cleaning robots are used in many households. Of the science fiction visions in which machines relieve humans of all mundane everyday tasks, they are the only widespread realization, apart from robotic lawnmowers.
Roomba is something like the iPhone of the industry: Introduced in 2002, the round robot was not the first of its kind. But it was the first to prevail. The Intelligent FloorVac for $ 200 defined the product category. To this day, Roomba defends its place despite many imitators. More than 20 million units have been sold, and there is no more successful home service robot.
Clara Vu played a significant part in this success. Her first job in 1998 after studying mathematics at Yale was to write the code that controlled Roomba. Its manufacturer iRobot, a start-up of three graduates from MIT’s elite university in Boston, was able to go public thanks to its success. Vu had enough and moved on.
She took her experiences with her. Today, she is the co-founder and chief technology officer of Veo Robotics. This Boston start-up doesn’t deal with household robots – and yet takes care of the interaction between humans and machines: Veo has developed a software platform that enables industrial robots and Enables people in production. These devices are called cobots.
Using sensors, the software creates a 3D map of the work area, allowing factory workers to work close to robots without getting in the way of one of the agile and heavy arms. Veos software is used in the construction of cars, airplanes, and refrigerators.
Substitute for wage labor
You can learn a lot about the robotics industry and its failed hopes on Vu’s path. The International Federation of Robotics (IFR) anticipates that worldwide sales of household robots will increase by 46 percent annually to $ 10 billion from 2018 to 2022. In 2018, however, the increase was only 20 percent – the breakthrough, it seems, is always imminent.
In contrast, it has long been successful in industry: in 2018, two and a half times as many industrial robots were installed as in 2013, 422,000 instead of 178,000. Where robots can replace wage workers, they are being used more frequently, retail robots, cleaning robots at airports, and delivery robots for parcels and pizzas on the streets of some US cities.
“When I left iRobot, it said there: We are not a vacuum cleaner manufacturer, but a robot manufacturer – so let’s find other jobs for robots in the household,” says Vu today. “But what are robots good at? They are strong, fast, and do a lot of repetitions.” Where these qualities are not needed, robots are often an expensive, superfluous solution. So she turned to industry.
Vu rolls her eyes when she hears the name Zume. The Mountain View company in Silicon Valley built robots that served pizzas. One who sprinkles tomato sauce on the dough, one who spreads it – and a six-axis robotic arm from ABB pushes the food into the oven.
Meanwhile, Zume pizza is a joke that is told in Silicon Valley – about the pre-corona decadence, in which even the stupidest ideas were amply financed. The pizza robot farmer received $ 375 million investment from the Softbank Vision Fund at the end of 2018, which later lost billions with investments in the failed office rental company WeWork or Uber.
Things are looking bad for Zume too: The start-up recently fired half of its employees and now wants to reinvent itself as a producer of face masks and sustainable packaging – the word pizza has been removed from the name. “People can do pizza pretty well themselves,” says Vu. “You don’t need a robot arm with six articulated joints.”
If you follow this logic, the robot as a human companion and helper is futuristic in everyday life, but a nonsensical idea in many areas of application.
No product has stood for the robot enthusiasm of recent years like Pepper, the friendly, white assistant with the tablet torso. When Softbank Robotics launched it in 2015, the first 1,000 copies were sold out in a minute. For a while, he rolled over the stage at every technology event or greeted guests in the atrium. According to its creators, it is said to “help people enjoy their lives”. However, the device developed in France never got beyond its status as a gimmick or information column.
High investments in hardware
Paolo Pirjanian also used to work at iRobot. When the Roomba manufacturer sucked in his company Evolution Robotics, which produced the mopping robot Mint, in 2012, the Iranian-born became technology director. Three years later, he left to start his own company: Embodied, which received investments from Amazon and Toyota and launched its first product a few months ago.
Moxie is a small, blue robot with big, green eyes that may have sprung from a Pixar film. It was designed by the Swiss Yves Behar, who worked for Apple, Samsung, and Prada and is considered one of the leading industrial designers in the world.
Moxie’s head is teardrop-shaped and, like the eyes, oversized – all according to the child’s scheme. It came on the market three months ago and is said to cost $ 1,500.
“We want to improve the emotional quotient of children,” says his inventor. Among those interested are schools, clinics, and parents of children with autistic behaviors. If Moxie becomes a success in the United States, Embodied plans to expand into other markets next year.
This is not a matter, of course. “I knew that robotics was a tough business for founders,” says the former refugee who grew up in Denmark and did his doctorate in Aalborg. You have to develop special hardware and software and electronics, which requires a lot of capital. When he wanted to raise money for Evolution in the 2008 financial crisis, “I was practically back on the aisle as soon as investors heard robotics,”
That has slowly changed in the past five years. Investors are more courageous today because some robot companies have, in the meantime, made large exits – Amazon’s acquisition of warehouse robot manufacturer Kiva Systems for $ 775 million has opened a door, even if it comes from the industry like most other success stories.
Pirjanian sees the mistakes in his industry: “Many consumer products are too expensive or their benefits are unclear,” says the founder. Some start-ups would have attracted a lot of venture capital and developed exciting products that would never have reached the mass market.
For example, Anki. The start-up from San Francisco, co-founded by Hanns Tappeiner from South Tyrol, raised $ 200 million to develop cute, tractor-shaped toy robots with Pixar faces. Although the company proudly stated in 2018 that it had sold more than 1.5 million robots since 2013, it went bankrupt in 2019. Or Jibo, a robot for families, started a crowdfunding campaign in 2014 and was discontinued three years later.
“The most obvious example is Pepper – a product with no clear purpose,” says Pirjanian. What differentiates him from Roomba. Before its market launch, the traditional company Electrolux had offered a vacuum robot for $ 2,000, but hardly sold any. “Three things are important for robots for private individuals,” says the founder. “The price, the price, and the price.”
Humanoids in the eerie valley
But how does Moxie go with it? A $ 1,500 robot for kids, isn’t that Anki 2.0, just more expensive? “Moxie is not a toy,” says Pirjanian. “If it were a toy, it would fail.” The drophead makes eye contact, recognizes facial expressions, and reacts to them. Moxie adds value by supporting a child’s development.
His big green eyes and mouth reflect his mood, and he is so intelligent that he can talk and play games with his somewhat tinny voice instead of just responding to a small number of commands. “It is a huge responsibility to give a robot a flexible mouth alone,” says Pirjanian. If the eyes and mouth are not precisely coordinated, a robot looks “scary and creepy”.
Numerous software and hardware problems had to be solved to program Moxie cute. A $ 1,500 robot that scares children would be the worst-case scenario for Pirjanian’s company.
The problem of making robots human-like even has a name: “Uncanny Valley.” The Japanese roboticist Masahiro Mori hypothesized that the acceptance of robots does not increase with their increasing human resemblance, but runs parabolically. Hence, it only has to go through a long valley.
Vacuum cleaners or industrial robots do not scare people because they have no humanoid appearance. Probably not perfect images of people because they would be authentic. On the other hand, robots that only rudimentarily mastered the human language, facial expressions, or gestures populated the eerie valley because, although they remind of people, they also disturb with their strange behavior.
“That’s why Pepper is not very interactive for an interactive robot,” says Pirjanian. The false eyes, communicating via a touchscreen – though all that prevents that the white Softbank -Robots his interlocutors give a sinking feeling. But it is also not particularly useful.
Although Vu and Pirjanian have worked on very different projects since their time at iRobot, they have one thing in common: They want to make robots useful and inspire confidence – whether it is a sizeable gripping arm or a small companion.