Last winter, NASA technicians sent a humanoid robot dubbed Robonaut 2 to the International Space Station. R2, which has only a torso, sophisticated arms and fingers, and a head full of sensors, was the result of a joint effort by NASA and General Motors to create a robot that could operate safely alongside humans. Robots like R2 could carry out dangerous or tedious tasks on space missions, but they’d also be useful on the ground, where they could assist factory workers. Robots have typically been segregated from humans for safety reasons, but improvements mean they’re now poised to take on a wider variety of tasks.
Although robots have aided manufacturing for decades, they’ve tended to be bulky systems that require precise setup to do large-scale, repetitious tasks such as welding or painting a car door. But improved technologies for vision processing and gripping are leading to a new wave of robots. In June, President Obama announced a $500 million federal investment in manufacturing technology (including $70 million for robotics), to partners that include Ford, Caterpillar, MIT, Carnegie Mellon University, and others. Though the partnership does not include the R2 project, it represents another step in developing robots that can assist with repetitious or physically stressful assembly-line tasks without posing a safety risk.
“In manufacturing facilities, robots are basically in cages like wild animals … so you can’t get in there and get hurt,” says David Bourne, a professor at Carnegie Mellon who works on robotic manufacturing. Having “the robot and person work side-by-side is really scary to a lot of people,” he says. “If it swings around and hits you, it could take your head off.”
R2 uses a popular robotics technology called series elastic actuators in its joints. The actuators have an elastic spring component between the motor and the object the robot has to pick up. The actuators help the robot detect and control the force of its own movements.
“The use of series elastic actuators changes the whole approach to manufacturing robots. [It] makes the robot able to safely interact with people,” says Rodney Brooks, a cofounder of iRobot and founder of Heartland Robotics, which is developing inexpensive, adaptable manufacturing robots and has licensed the series elastic actuator.
In addition to its force-sensing joints, R2 is covered in soft material in case of accidental collisions, and its head is chock-full of cameras—including an infrared camera for depth sensing—so it can keep track of its human colleagues.
“The fact that Robonaut 2 is on the space station is a great example of how safe it is,” adds Brooks. “It’s very promising for assembly operations.”
In addition to its safety features, R2 is more versatile than traditional factory robots at gripping things. Each of its dexterous, humanlike fingers can hold up to five pounds, and the arm can hold around 20 pounds in a variety of positions: the R2 can dial an iPhone and lift a barbell. The series elastic actuators allow R2 to feel the force of objects, rather than only calculate their position. This is how a human does it: when we snap a battery backing onto a phone, for example, we’re guided by the feel of the forces we exert on the phone.
“Traditionally, industrial robots have to do things by very precise position control. Once you have force control, you can be less precise and operate like a human. Then it’s much easier to program the robot to do assembly,” says Brooks. Marty Linn, principal robotic engineer for GM, predicts that technology from the R2 hand and arm will likely be part of the next wave of assembly-line robots. The versatility of its grip means the robot could manipulate many different objects with the same hand, which would make it faster at adapting to a new task. “We really want our assembly processes to be very flexible,” Linn says.
GM uses about 2,500 new robots every year, and has around 20,000 to 25,000 robots in factories worldwide, according to Linn. The new generation of robots would not replace these cumbersome giants, but do smaller, more sophisticated tasks, such as handling the screws, handles, and airbag and blind-spot warning sensors that go into the car doors. That kind of work is “ergonomically challenging” for humans, Linn says.
However, Bourne says, it might be 20 years before we see robots working with people, because it will take that long to create and verify safety standards. “The barrier of entry is mostly in the liability,” he says.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
These materials were meant to revolutionize the solar industry. Why hasn’t it happened?
Perovskites are promising, but real-world conditions have held them back.
Why China is still obsessed with disinfecting everything
Most public health bodies dealing with covid have long since moved on from the idea of surface transmission. China’s didn’t—and that helps it control the narrative about the disease’s origins and danger.
A quick guide to the most important AI law you’ve never heard of
The European Union is planning new legislation aimed at curbing the worst harms associated with artificial intelligence.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.