MIT scientists present how they’re growing AI for humanoid robots | USAEMALL.com

We’ve all seen what artificial intelligence can do on our screens: generate art, carry out conversations and help with written tasks. Soon, AI will be doing more in the physical world.

Gartner, a research and advisory firm, estimates that by 2030, 80% of Americans will interact daily — in some way — with autonomous, AI-powered robots. 

At the Massachusetts Institute of Technology, professor Daniela Rus is working to make that possible — and safe.

“I like to think about AI and robots as giving people superpowers,” said Rus, who leads MIT’s Computer Science and Artificial Intelligence Lab. “With AI, we get cognitive superpowers.”

“So think about getting speed, knowledge, insight, creativity, foresight,” she said. “On the physical side, we can use machines to extend our reach, to refine our precision, to amplify our strengths.”

Sci-fi stories make robots seem capable of anything. But researchers are actually still figuring out the artificial brains that machines need to navigate the physical world. 

“It’s not so hard to get the robot to do a task once,” Rus said. “But to get that robot to do the task repeatedly in human-centered environments, where things change around the robot all the time, that is very hard.”

Rus and her students have trained Ruby, a humanoid robot, to do basic tasks like prepare a drink in the kitchen. 

“We collect data from how humans do the tasks,” Rus said. “We are then able to teach machines how to do those tasks in a human-like fashion.”

Rus’ students wear sensors to capture motion and force, which helps teach robots how tightly to grip or how fast to move.

“So you can tell, like, how tense they’re holding something or how stiff their arms are,” said Joseph DelPreto, one of Rus’ students. “And you can get a sense of the forces involved in these physical tasks that we’re trying to learn.”

“This is where delicate versus strong gets learned,” Rus said.

Robots already in use are often limited in scope. Those found in industrial settings perform the same tasks repeatedly, said Rus, who wants to expand what robots can do.

One prototype in her lab features a robotic arm that could be used, in the future, for household chores or in medical settings. 

Some, however, might feel uneasy having robots in home settings. But Rus said every machine they’ve built includes a red button that can stop it. 

“AI and robots are tools. They are tools created by the people for the people. And like any other tools they’re not inherently good or bad,” she said. “They are what we choose to do with them. And I believe we can choose to do extraordinary things.”