Skip to Content

Cheaper Joints and Digits Bring the Robot Revolution Closer

Efforts to build robot hands and humanoids more cheaply could make them affordable enough for businesses and even homes.
April 4, 2014

The Atlas humanoid robot, unveiled last year by Boston Dynamics, a company later acquired by Google, is a marvel. It can clamber over rubble and operate power tools. But these abilities don’t come cheap. Atlas has a price tag well above a million dollars, and it consumes around 15 kilowatts of electricity when in operation, meaning hefty power bills for its owner and limiting its practicality. “That’s enough to power a small city block,” says Alexander Kernbaum, research engineer at the nonprofit research agency SRI International. To be truly practical, he says, Atlas “needs to be many times more efficient.”

electrostatic clutch
Helping hand: A new kind of electrostatic clutch from SRI makes this design around 10 times cheaper than previous robotic hands, which could cost $35,000 or more.

Kernbaum is part of a team at SRI that recently began working on that problem under a contract with DARPA, the Pentagon research agency (Atlas itself was built with DARPA funding). The team aims to rethink the robot’s design to preserve its capabilities but slash its power usage by at least 20 times, putting it on par with a microwave oven.

SRI won’t talk about how that will be done. But the general approach will be to replace the power-hungry hydraulics that move Atlas’s joints with a smaller number of lighter, more efficient, and cheaper electric components that can achieve the same thing.

Rethinking the components used in advanced prototypes such as Atlas to reduce cost and power consumption has become a major focus in robotics research as engineers seek to finally have these machines escape the lab, says Rich Mahoney, SRI’s director of robotics. “We got things that are overdesigned because there’s not been impetus for low cost and good design,” he says.

For a long time researchers have been focused on simply answering basic questions of whether functioning, agile humanoids could be built, says Mahoney. “We were in the domain of ‘Is this possible?’ ” He says this question has now been answered, so the time is right to drive down the costs of the components used in sophisticated robot legs, arms, and hands, making them affordable to small businesses and even consumers. “Manipulation is simply not available at that level now,” says Mahoney. “But it can be.” He says cheaper components would make it possible for humanoids like Atlas to become standard safety tools in places like oil rigs. “Instead of ‘In case of emergency break glass,’ and there’s a hatchet, there would be a humanoid.”

three-fingered hand made by iRobot
Nimble fingers: This three-fingered hand made by iRobot can use its nails to pick up small objects. It could be made for around $3,000 or less.

More immediately, these advances could help a market that Melonee Wise, CEO and cofounder of Unbounded Robotics, calls service robotics. “It’s when you start looking at the robot doing human-scale tasks,” she says. “That means having to sense and manipulate in complex ways in a complex workspace, like moving cans in a refrigerator.”

The poster child of this kind of robotics is Baxter, a robot with two arms and simple grippers that can work in factories alongside humans (see “This Robot Could Transform Manufacturing”). Wise’s company makes a $35,000 mobile robot, called the URB-1, that has a single arm. Like Baxter, it is intended to work alongside people in warehouses and other human workspaces (see “This Might Be the Model T of Workplace Robots”). Both Baxter and URB-1 can do a lot with simple grippers and today’s arm technology, says Wise. But being able to make use of lower cost or more capable technology would be a major boost to robots intended for human workspaces.

Several low-cost robotic hands recently emerged from another DARPA program called ARM-H. By achieving greater complexity at lower costs, these hands could help Baxter or Unbounded’s robots perform new tasks. Roomba manufacturer iRobot worked with Harvard and Yale to create a three-fingered hand that can do anything from holding a basketball to picking up a key lying flat on a table.

If it were made in quantities of a few thousand, the hand should cost around $3,000, says Mark Claffee, principal robotics engineer at iRobot, which also makes military and telepresence robots. “It’s a dramatic change,” Claffee says, as the current going rate for a robotic hand with similar capabilities starts at around $35,000.

One way iRobot cut costs was to use rubber in the joints of the hand, introducing springiness that allows it to get a good grip on something without specifying the position of every finger exactly for different objects. Costs were also slashed by giving the hand only three fingers. It’s designed so that three digits together can grasp something, and two can be opposed to manipulate smaller objects.

“We believe that manipulation is going to be a game changer,” says Claffee. “Imagine having a telepresence platform like our Ava 500 that can one day pick things up.” He is now working to apply some of the same design techniques used in the finger to robot arms.

Claffee says the work will allow robots to function in much messier environments than Baxter. “Those robots are capable but expensive and they’re focused on industrial applications,” says Claffee. “To reach consumers, we need to significantly reduce costs, and have manipulation for human environments that are much less structured.”

SRI also took part in the ARM-H program and made its own low-cost robotic hand. This hand makes use of a cheaper, lighter, and more efficient clutch mechanism; it uses electrostatic forces to lock a joint in place. Making widespread use of clutches allows a single motor to control all three joints in a finger.

“This could lead to manipulation solutions under a thousand dollars, which makes this kind of robotics more aligned with the consumer domain,” says Mahoney, pointing out that small aerial and ground robots are already available to consumers, thanks in part to the smartphone boom. If Mahoney and others are right, cheap but capable robot hands and limbs may be next to become affordable.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.