Controls a Robot Arm Simply by Thinking
The Korea Advanced Institute of Science and Technology (KAIST) announced on Feb. 23 that its research team led by professor Jung Jae-sung has developed a brain-machine interface system for controlling a robot arm simply by thinking.
According to the team, the system employs artificial intelligence and genetic algorithms and catches arm movement-related intentions simply with brain waves measured in the deep cerebrum. “Our brain-machine interface model is capable of moving in 24 directions in three dimensions, that is, in eight directions in each dimension,” it said, adding, “Its accuracy is between 90.9 percent and 92.6 percent in every direction.”
The team used reservoir computing in its research so that AI-based learning is possible even with low-spec hardware.
“Most existing brain-machine interface systems require high-spec hardware, which means limited real-time application and limited application to smart devices,” the professor said, continuing, “Our development can be widely used in metaverse environments and smart devices for avatar control, app control, and so on.”
Details of the research have been published in the March edition of the Applied Soft Computing journal.