Research team led by professors Gong Kyoung-chul and Kim Jung of the mechanical engineering department embarks on bidirectional Brain-to-Robot development

A schematic diagram of the bidirectional brain-robot interface. [Provided by KAIST]
A schematic diagram of the bidirectional brain-robot interface. [Provided by KAIST]

Development has begun on a two-way brain-robot interface that allows a person to control an exoskeleton through thought alone while simultaneously receiving tactile and force information sensed by the robot back into the brain.

A research team led by professors Gong Kyoung-chul and Kim Jung of the KAIST Department of Mechanical Engineering announced Thursday that it has launched development of the world's first bidirectional "Brain-to-Robot" system in partnership with Angel Robotics, as a flagship project under the government's cross-ministry advanced medical device research and development program.

Brain interface technologies that use neural signals to move a cursor or control a smartphone have already entered human clinical trials, with global companies such as Neuralink and Synchron in the United States accelerating their own development efforts. Existing technologies, however, have been limited in their ability to link real movement and sensation simultaneously. The field has largely focused on advancing signal-decoding techniques without clearly defining what the brain signals actually control or what sensory feedback is returned to the user.

In the Brain-to-Robot system, AI interprets a user's brain signals to identify movement intent and converts them into robot control commands. At the same time, tactile, pressure and force information detected by the robot's skin and sensors is converted into brain stimulation signals and delivered back to the user.

An integrated architecture for a bidirectional brain-robot interface designed to restore function in people with lower-limb paralysis. [Provided by KAIST]
An integrated architecture for a bidirectional brain-robot interface designed to restore function in people with lower-limb paralysis. [Provided by KAIST]

Professor Gong's team will develop wearable robot control and AI-based movement-intent interpretation technologies, and design a somatosensory interface — a system for transmitting bodily sensations — to accurately relay sensory information detected by the robot to a Brain Chip, a semiconductor that processes neural signals. Professor Kim's team will develop robotic skin capable of sensing on behalf of people with disabilities, along with AI-based somatosensory interpretation technology.

The project will also pursue AI-based encoding and decoding algorithms that convert brain signals into robot control commands and relay sensory information back to the brain. A core technical challenge will be processing cortical signals — neural signals generated in the cerebral cortex — across hundreds of channels in real time while maintaining an extremely low-latency closed loop, a circular control structure in which signals are continuously exchanged in real time.

Angel Robotics, the company founded by Professor Gong, will handle commercialization, with plans to pursue full-cycle market entry covering everything from Ministry of Food and Drug Safety approval to actual distribution.

"There is potential for this technology to be used in rehabilitation and daily life support for patients with quadriplegia, spinal cord injuries and severe motor disorders," Professor Gong said. "Our long-term goal is to build a rehabilitation platform that enables patients to walk on their own, pick up objects and even feel the sensation of touch."


nbgkoo@heraldcorp.com