Getting robots to do things would be a whole lot easier if people could command them to perform tasks using their minds.
That dream is now a reality. With a new robot control system, a human can control a robot using nothing but brainwaves and hand gestures.
MIT Uses Brain Signals, Hand Gestures To Control Robots
The new technology, which is developed at MIT's Computer Science and Artificial Intelligence Laboratory, allows users to control a robotic arm more intuitively, using brain signals and hand movements. This enables the operator to instantly correct mistakes made by the robot with their mind or a simple flick of a finger.
Being able to control robots in this way opens up new possibilities for how humans could manage teams of robot workers or machines. This robot control system, which will be presented at the Robotics: Science and Systems conference in Pittsburgh later this month, could be useful to people who oversee robots in factories, homes, or hospitals to make sure bots operate in a safe and efficient manner.
How Does It Work?
The technology harnesses the power of brain signals called "error-related potentials" or ErrPs that are automatically generated when humans notice a mistake via electrodes worn on the operator's head and forearm.
The system monitors the brain activity of the operator and if an ErrPs occurs due to an error made by the robot, it sends an alert to the robot. As soon as the bot receives the signal, it stops what it is doing.
The operator can then control the robot and correct it using hand gestures. These movements are detected by the electrodes worn on the operator's arm that monitor muscle activity.
Demonstration With 'Baxter'
The researchers demonstrated the technology with the help of "Baxter," a robot from Boston-based Rethink Robotics. Baxter was equipped with a power drill and given the task to drill one of three possible targets on the body of a mock-up plane fuselage.
Baxter would select a target and go for it. If the operator noticed that the humanoid had chosen the incorrect target, he would simply think about it in order to stop Baxter. He would then use hand gestures to tell the robot which holes to drill.
With human supervision, the robot's target accuracy increased from 70 percent to 97 percent. Unlike traditional robotic management, Baxter learns from the way the operator thinks instead of the user having to learn the robot's language.
"What's great about this approach is that there's no need to train users to think in a prescribed way," said Joseph DelPreto, the project's lead author. "The machine adapts to you, and not the other way around."