“A Neurally Controlled Robotic Arm”

May 16th, 2012 by Erwin Gianchandani Post a comment »

An article published in Nature this afternoon is generating some buzz in the news media, largely for the advances in fundamental computing research, particularly with respect to brain-machine interfaces. From a msnbc.com story:

In a clinical trial, Cathy Hutchinson, who has been paralyzed for more than 14 years, used the BrainGate system to mentally control a robotic arm and reach for a drink [image courtesy The BrainGate Collaboration via msnbc.com].The stroke that disconnected Cathy Hutchinson’s brain from her body has kept her silent and unable to move for more than 14 years. But science is starting to change all that.

 

Researchers have connected the 58-year-old woman’s brain to a computer that runs a robotic arm. As Hutchinson sits at a table staring at a bottled drink and imagining the robot grabbing the bottle and bringing it to her mouth, the robot arm begins to move.

 

The robot is running on signals detected by sensors implanted in the part of Hutchinson’s brain that would normally control the movements of her right arm. The sensors pick up the sparking of nerve cells and send the signals to the computer which then translates them into commands for the robotic arm. Suddenly Hutchinson is able to do something she could only dream of before: As she thinks about getting herself a drink, the arm reaches over to the bottle and brings it to her lips, where she is able to sip the drink from a straw.

 

It’s the first time Hutchinson’s been able to do anything for herself since the stroke [more after the jump].

 

 

It will be years before BrainGate could be available to the general public. But Hutchinson’s happy to enjoy the future today.

 

 

What’s amazing is how researchers have “taught” their computer to essentially read Hutchinson’s thoughts.

 

The baby-aspirin sized sensor implanted in Hutchinson’s brain contains 96 hair-thin electrodes that record the sparking of neurons in the movement control center, the motor cortex.

 

The first step in the learning process is for the computer to “see” which neurons spark, and in what pattern, when a person picks up a bottle and brings it to her lips, explains the study’s lead researcher, Dr. Leigh R. Hochberg, a professor of engineering at Brown University, a researcher at the Providence VA Medical Center, a critical care neurologist at the Massachusetts General Hospital/Brigham and Women’s Hospital in Boston, and a visiting associate  professor of neurology at the Harvard Medical School.

 

Fortuitously, it doesn’t matter whether the person actually moves their limb or whether they’’e merely imagining themselves doing it. So, for several trials, Hochberg and his colleagues had the computer observe the sparking patterns of neurons in Hutchinson’s brain as she watched the robot arm pick up the bottle and bring it to her lips.

 

 

Once the scientists had taught the computer which patterns would normally make Hutchinson’s arm reach out for the bottle of coffee, they hardwired them as the command for the robot arm to do the same thing – but with the signal coming directly from Hutchinson’s brain as she imagined herself grasping the cup and bringing it to her lips.

 

“Beyond this, our real dream for this research is for people with paralysis — from a brain stroke or spinal cord injury — to be able to one day reconnect the brain to the limbs,” Hochberg says…

Read the full msnbc.com story — and see the research article (subscription required) in today’s issue of Nature.

(Contributed by Erwin Gianchandani, CCC Director)