Trick Hat Media
A robot that uses AI to learn tasks like folding laundry
Blue Robot

UC Berkeley has designed a robot named ‘Blue’ that can use AI to learn basic tasks like folding laundry through trial & error. The “low-cost”, “human-safe” robot can also learn from demonstrations by a human controller. “With a lower-cost robot, every researcher could have their own robot…that vision is one of the main driving forces behind this,” a researcher said.


A new low-cost, human-friendly robot uses recent advances in artificial intelligence & deep reinforcement learning to master intricate human tasks.

Robots may have a knack for super-human strength & precision, but they still struggle with some basic human tasks—like folding laundry or making a cup of coffee.

That’s where the new robot, named Blue, comes in. Researchers have designed Blue to be capable while remaining affordable & safe enough that every artificial intelligence researcher—& eventually every home—could have one.

Blue is the brainchild of Pieter Abbeel, a professor of electrical engineering & computer sciences at the University of California, Berkeley; postdoctoral research fellow Stephen McKinley; & graduate student David Gealy. The team hopes Blue will accelerate the development of robotics for the home.

A ROBOT FOR THE AI MOMENT

“AI has done a lot for existing robots, but we wanted to design a robot that is right for AI,” Abbeel says.

“Existing robots are too expensive, not safe around humans, & similarly not safe around themselves—if they learn through trial & error, they will easily break themselves. We wanted to create a new robot that is right for the AI age rather than for the high-precision, sub-millimeter, factory automation age.”

Over the past 10 years, Abbeel has pioneered deep reinforcement learning algorithms that help robots learn by trial & error or through human guidance like a puppet. He developed these algorithms using robots from outside companies, which market them for tens of thousands of dollars.

Blue’s durable, plastic parts & high-performance motors total less than $5,000 to manufacture & assemble. Its arms, each about the size of the average bodybuilder’s, are sensitive to outside forces—like a hand pushing it away—& has rounded edges & minimal pinch points to avoid catching stray fingers. Blue’s arms can be very stiff, like a human flexing, or very flexible, like a human relaxing, or anything in between.

“With a lower-cost robot, every researcher could have their own robot, & that vision is one of the main driving forces behind this project—getting more research done by having more robots in the world,” McKinley says.

GOING WITH THE FLOW

Robotics has traditionally focused on industrial applications, where robots need strength & precision to carry out repetitive tasks perfectly every time. These robots flourish in highly structured, predictable environments—a far cry from the traditional American home, where you might find children, pets, & dirty laundry on the floor.

“We’ve often described these industrial robots as moving statues,” Gealy says. “They are very rigid, meant to go from point A to point B & back to point A perfectly. But if you command them to go a centimeter past a table or a wall, they are going to smash into the wall & lock up, break themselves, or break the wall. Nothing good.”

If an AI is going to make mistakes & learn by doing in unstructured environments, these rigid robots just won’t work. To make experimentation safer, Blue was designed to be force-controlled—highly sensitive to outside forces, always modulating the amount of force it exerts at any given time.

“One of the things that’s really cool about the design of this robot is that we can make it force-sensitive, nice & reactive, or we can choose to have it be very strong & very rigid,” Gealy says.

“Researchers can adjust how stiff the robot is, & what kind of stiffness—do you want it to feel like molasses? Do you want it to feel like a spring? A combination of those? If we want robots to move toward the home & perform in these increasingly unstructured environments, they are going to need that capability.”

To achieve these capabilities at low cost, the team considered what features Blue needed to complete human-centered tasks, & what it could go without. For example, the researchers gave Blue a wide range of motion—it has joints that can move in the same directions as a human shoulder, elbow, & wrist—to enable humans to more easily teach it how to complete tricky maneuvers using virtual reality. But the agile robot arms lack some of the strength and precision of a typical robot.

“What we realized was that you don’t need a robot that exerts a specific force for all time, or a specific accuracy for all time. With a little intelligence, you can relax those requirements & allow the robot to behave more like a human being to achieve those tasks,” McKinley says.

WEAKER BOT, MORE POWER

Blue is able to continually hold up 2 kilograms [just under 4.5 pounds] of weight with arms fully extended. But unlike traditional robot designs that are characterized by one consistent “force/current limit,” Blue is designed to be “thermally-limited,” McKinley says.

That means that, similar to a human being, it can exert a force well beyond 2 kilograms in a quick burst, until it reaches its thermal limits & it needs time to rest or cool down. This is just like how a human can pick up a laundry basket & easily carry it across a room, but might not be able to carry the same laundry basket over a mile without frequent breaks.

“Essentially, we can get more out of a weaker robot,” Gealy says. “And a weaker robot is just safer. The strongest robot is most dangerous. We wanted to design the weakest robot that could still do really useful stuff.”

“Researchers had been developing AI for existing hardware & , about three years ago, we began thinking, ‘Maybe we could do something the other way around. Maybe we could think about what hardware we could build to augment AI & work on those two paths together, at the same time,’” McKinley says. “And I think that is a really dramatic shift from the way a lot of research has taken place.”

Currently, the team is building 10 arms in-house to distribute to select early adopters. They are continuing to investigate Blue’s durability & to tackle the formidable challenge of manufacturing the robot on a larger scale, which will happen through the UC Berkeley spinoff Berkeley Open Arms, where sign-ups for expressing interest in priority access are now open.

Source: UC Berkeley