This AI-based device has been running continuously for 7 months without needing any adjustment, compared to similar devices that only work for 1 or 2 days.
Neuroscience professor Karunesh Ganguly (center) and his mind-controlled robotic arm - Photo: ucsf
Researchers at the University of California, San Francisco, USA have helped a paralyzed man control a robotic arm through a device that transmits signals from his brain to a computer. He can grasp, move and release objects just by imagining himself performing those movements.
The device, called a brain-computer interface (BCI), has been operating continuously for seven months without needing adjustment — a remarkable record, given that similar devices have previously only worked for a day or two.
The BCI relies on an artificial intelligence (AI) model that can adapt to small changes in the brain as a person repeats a movement — or in this case, an imagined movement — and learns to perform it more subtly.
“The combination of human and AI learning is the next step in brain-computer interfaces. This is essential to achieve sophisticated, near-real-world functionality,” said neuroscientist Karunesh Ganguly.
The study, funded by the National Institutes of Health and published in the journal Cell on March 6, was key in finding changes in brain activity from day to day as participants repeatedly imagined performing specific actions. Once the AI was programmed to adapt to these changes, it could work for months at a time.
The team worked with a participant who had been paralyzed by a stroke years earlier and was unable to speak or move. Tiny sensors implanted on the surface of his brain were able to record brain activity as he imagined movement.
After practicing with the virtual robotic arm, the participant was able to control the real robotic arm to pick up wooden blocks, rotate them, and move them to new locations. He was even able to open a cabinet, grab a cup, and bring it to the water dispenser.
For paralyzed people, the ability to feed themselves would be a huge life change. Professor Ganguly believes this is entirely possible in the near future. He is continuing to perfect the AI model to help the robotic arm move faster and smoother, and plans to test the BCI in a home environment.
Source: https://tuoitre.vn/chi-dung-y-nghi-nguoi-dan-ong-bi-liet-co-the-mo-tu-rot-nuoc-bang-tay-robot-20250309153210475.htm
Comment (0)