Controlling Robots With Your Thoughts

Over the last few months, in José del R. Millán’s computer science lab in at the École Polytechnique Fédérale de Lausanne, a little round robot, similar to a Roomba 
with a laptop mounted on it, bumped its way through an office space filled with furniture and 
people. Nothing special, except the robot was being controlled from a clinic more than 60 miles 
away—and not with a joystick or keyboard, but with the brain waves of a paralyzed patient.


The robot’s journey was an experiment in shared control, a type of brain-machine interface that merges conscious thought and algorithms to give disabled patients finer mental control over devices that help them communicate or retrieve objects. If the user experiences a mental misfire, Millán’s software can step in to help. Instead of crashing down the stairs, for instance, the robot would recalculate to find the door.

Such technology is a potential life changer for the tens of thousands of people suffering from locked-in syndrome, a type of paralysis that leaves patients with only the ability to blink. The condition is usually incurable, but Millán’s research could make it more bearable, allowing patients to engage the world through a robotic proxy. “The last 10 years have been like a proof of concept,” says Justin Sanchez, director of the Neuro­prosthetics Research Group at the University of Miami, who is also studying shared control. “But the research is moving fast. Now there is a big push to get these devices to people who need them for everyday life.”


Millán’s system, is a big step in making brain-machine interfaces more useful by splitting the cognitive workload between the patient and the machine. Previously, users had to fully concentrate on one of three commands—turn left, turn right, or do nothing—
creating specific brain wave patterns detected by an electrode-studded cap. That system exhausted users by forcing them to think of the command constantly. With shared control, a robot quickly interprets the user’s intention, allowing him to relax mentally. Millán is now developing software that is even better at weeding out unrelated thoughts and determining what the user really wants from the machine.


Discover