🦾 A Paralyzed Man Learned to Mind-Control a Robotic Hand
Researchers from the University of California, San Francisco, have created an AI-powered brain-computer interface (BCI) that enables a person who has been paralyzed for many years after a stroke to control a robotic arm.
Most such devices work properly for only a couple of days. Human brain activity patterns change daily, and the algorithms in BCI devices needed retraining as they were unable to recognize these changing patterns.
The new brain-computer interface automatically adjusts to these changes, and the model worked for seven months. Months later, the participant could still control the robotic arm after a 15-minute "tune-up" to adapt to the shifts in his brain activity since he had started using the device.
The experiment had several steps to follow. For the first two weeks, the fully paralyzed patient repeatedly imagined himself moving his fingers and toes, arms, and head. Tiny sensors implanted on the surface of his brain recorded his mental activity to train the AI.
Next, the patient learned to control a virtual robot arm, receiving feedback on the accuracy of his visualizations. After that, he began practicing with a real robot arm connected to the interface. He made the robotic arm pick up and move blocks, open a cabinet, grab a cup, and hold it under a water dispenser ⤴️
The study authors regard this new technology as a significant breakthrough in developing neuroprostheses, offering paralyzed individuals the potential to perform everyday tasks independently. "This blending of learning between humans and AI is the next phase for these brain-computer interfaces," said Karunesh Ganguly, a professor of neurology at the University of California, San Francisco.
More on the topic:
🔄 Japanese Scientists Created a Biohybrid Robotic Hand
