Patients with paralysis control robot arms with thought

Research team at the University of California San Francisco, USA, helping a man with paralysis control robot arms through brain -to -brain transition devices, computer to computer, Interesting Engineering On 6/3 reported. He can hold, move and drop objects just by imagining himself to do those actions. New research has been published in the magazine Cell.

The new device, called the brain -computer interface (BCI), works for a record 7 -month record without much adjustment. Previously, similar devices could only operate in one or two days.

The biggest progress comes from the artificial intelligence model (AI) that this BCI is built based on. It adapts to natural changes in brain activity over time, allowing participants to tweak their imagination movements.

“The combination of learning between people and who is the next stage of brain -computer interfaces. That’s what we need to achieve sophisticated, like real functions,” said Karunesh Gangule neuroscientist.

The man participated in the study, was paralyzed by stroke, transplanted small sensors on the brain surface. When this person imagines the movement of limbs or heads, the sensor will record brain activity. The team found that, although the brain movement samples consistently in shape, their exact position changed slightly from day to day. This helps explain why the previous BCI failed.

To solve this problem, the research team develops a AI model that can be adjusted according to daily changes. For 2 weeks, the patient tries to visualize simple movements while who learns from brain signals. When this person first tried to control the robotic arm, the movement was not accurate. In order to improve the accuracy, he practiced with a virtual robot arm that provides feedback in real time.

As soon as you learn how to use virtual arms, the patient quickly applies these skills to the real robotic arm. He can hold the box blocks with robots, rotate and place them in many different positions. In a more complex task, he opened the closet, took out a cup and placed it under the faucet of the tree. Many months later, the patient can still use the robotic arm after only a short adjustment of about 15 minutes.

Gangule and colleagues are focusing on improving AI to move the robot’s arm to become smoother and faster. They also aim to test the system in the actual family environment. “I believe we have grasped the system to build the system and can make it effective,” Gangully said.