Quadriplegic successfully uses mind-controlled robotic arm
Back in February 2012, Schuermann had two electrode grids – with 96 contact points each – surgically implanted into her left motor cortex, which controls right arm and hand movements. Hooked up to a brain-computer-interface, she was able to reach in and out, left and right, and up and down with the arm within a week of the surgery. In a few months she learned to flex the wrist back and forth, move it from side to side, and rotate it. She could also grip objects using a crab-like pincer hand shape.
For this second study, the researchers sought to increase the hand’s functional utility while also maintaining a relatively low-complexity control space because more possible maneuvers means considerably more calibration. The 10-dimensional (10D) control (three dimensions of translation, three of orientation, four of hand shape) that they settled on had not been tested elsewhere, and they were unsure whether the brain would exhibit a preference for just one dimension or simultaneously handle all 10.
To calibrate the arm, Scheuermann watched animations of the movements and new hand shapes and imagined herself doing them while the team recorded the resultant signals in her brain. After identifying the brain activity patterns that correspond to each movement, these signals were then translated into instructions for the robotic arm.
The researchers also note with curiosity that Schuermann found objects easier to grasp if they had been displayed during the preceding calibration.
Schuermann underwent surgery in October to have the electrode arrays removed. She speaks of the experience fondly, saying “this study has enriched my life, given me new friends and coworkers, helped me contribute to research, and taken my breath away.”
For other victims of paralysis, the research provides a new hope. “Our results indicate that highly coordinated, natural movement can be restored to people whose arms and hands are paralyzed,” said co-author Andrew Schwartz.
A paper describing the research was published in the Journal of Neural Engineering.