According to Melonee Wise, the manual laborer of the future has only one arm and stands just three feet, two inches tall. Such are the vital statistics of UBR1, a $35,000 mobile robot unveiled today by Wise’s startup company Unbounded Robotics.
Wise, the company’s CEO and cofounder, says her business will at first sell the robot to researchers in academia and industry, who currently must either pay hundreds of thousands of dollars to get hold of a similar robot or build one themselves. But the UBR1 has also been designed to be capable and safe enough to help out in real workplaces such as warehouses and factories.
Unbounded will start shipping the first UBR1s next summer. Within the next two years, Wise expects to see businesses start putting them to work at tasks such as bin picking or stocking shelves in warehouses. “We’re in the business of getting robots out of the lab and making them a platform for businesses to use,” she says. “This is the Model T of robots.”
Though robots have long been a part of manufacturing, they have traditionally worked in isolation. But in recent years, thanks to advances in hardware and software, new kinds of robot have begun to appear among human workers in factories and warehouses. A company called Kiva Systems, acquired by Amazon in 2012, makes robots that can haul items around (see “In Warehouses, Kiva’s Robots Do the Heavy Lifting”), while startup Rethink Robotics’ flagship two-armed robot can work alongside humans on a production line (see “Baxter: The Blue-Collar Robot”).
These systems have limitations, though. Kiva’s robots require dedicated support infrastructure to be installed, and although the $22,000 Baxter is capable of two-handed manipulation, it cannot move around. Mobile robots capable of manipulation could function more like real human workers, Wise says, but so far they have remained in the research lab.
Before the UBR1 can start work in the real world, Unbounded will have to significantly improve the software available for the robot, which is today essentially a blank slate that requires a buyer to program in the desired capabilities. By contrast, Baxter is configured to be able to learn some manipulation tasks out of the box. Wise says her company will develop modular software packages that UBR1 owners can download to give their robots practical abilities. “We see ourselves developing basic capabilities that people will download and use on their robots, such as ‘open a door’ and ‘pick up a cup,’” she says.
UBR1 has a vaguely anthropomorphic head and a 75.75-centimeter (29.75-inch) arm, with four joints and a pincer grip, that neatly folds against its body when not in use. The robot moves around on a circular base with hidden wheels and can go anywhere in a building suitable for wheelchairs. The robot’s torso can be extended so that its head is anywhere between 96.5 centimeters (three feet, two inches) and 132 centimeters (four feet, four inches) from the ground. UBR1’s “eyes” actually hide the three lenses of a depth camera made by PrimeSense, a company that also supplies the hardware for Microsoft’s Kinect Xbox.
The robot also has design features that make it safe for work around people. For example, if it has accidentally shoved a person or invaded someone’s space, its jointed arm can be pushed aside.
Unbounded Robotics was spun out of the for-profit research lab Willow Garage this January by Wise and three fellow robotics engineers. Officials decline to talk about how the company is funded. The UBR1 makes use of the open-source Robot Operating System originally developed at Willow (see “TR35: Morgan Quigley”) and can be described as a simpler version of Willow’s flagship PR2, a large mobile robot with two arms that sold to research labs for $400,000. Although the PR2 became the basis for projects that pushed the boundaries of robot autonomy (see “Robots That Learn From People” and “TR35: Leila Takayama”), the high price meant that only a handful were sold. Wise says that just 43 PR2s exist in labs around the world today.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.