Watch This Robot Correct Its Mistakes Through Brain Waves


We all know humans are mistake-prone. But even robots mess up sometimes. A new collaboration between MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) and Boston University is studying how human brain signals can tell a robot to do the right thing, in real-time.

Remember Rosie, the robotic maid from "The Jetsons"? Imagine George Jetson simply thinking that she needs to load the dishes into the dishwasher when she's mistakenly heading to the washing machine.

The team didn't have Rosie, so they called on Baxter, a two-arm collaborative robot (one that's designed to work with people on different tasks). He's trained in industrial automation, so he packages, loads and unloads and handles materials. In this case, Baxter's job was to complete simple binary-choice, object-sorting tasks, like picking up spray-paint cans and putting them in the correct bucket.

Using machine-learning algorithms, the team developed a system that classifies brain waves in 10–30 milliseconds. That's much faster than tapping one letter on a keypad or pushing a button. To get Baxter to respond to people's brain waves, the research team hooked participants up to an electroencephalography (EEG) monitor to track brain activity.

In the past, EEG-controlled robotics required extensive, almost daunting, training processes so humans would think in a way that computers would recognize. This team sought a more natural approach by targeting the brain's "error-related potentials" (ErrPs), signals that humans produce when our brain notices mistakes. When a change in ErrPs occurs, the robot picks up on his mistake and adjusts. "As you watch the robot, all you have to do is mentally agree or disagree with what it is doing," says Daniela Rus, CSAIL director, in a press release. Just in case the robot gets a bit confused when completing a task, it can call upon a human response.

The researchers say the technology eventually could extend to multiple-choice tasks, or even help people who need assistance communicating verbally. So, advanced human-robot collaboration may not be as futuristic as "The Jetsons" had us believe.



More to Explore

#}