Brain-Computer Interface (BCI): Brain-Computer Connection Technology
Brain-Computer Interface (BCI) refers to a technology for controlling a computer or external device by measuring and analyzing neural signals generated from the cranial nervous system, or for transmitting the user's intentions and intentions to the outside.
BCI study for external device control
Core Technology
Noise and motion artifact removal technology
Development of BCI classification algorithm for Motor Imagery
Development of motion image BCI training technology
Development of adaptive BCI technology considering user status
Hybrid BCI technology development
Error recognition detection technology development
Application technology
Exoskeleton robot control
remote robot control
Meal assistant robot control
Research to improve BCI performance - brain signal preprocessing technology
Artifact Subspace Reconstruction (ASR)-based brain signal preprocessing algorithm is robust against various motions and noises. However, it requires a large number of EEG channels and require the individual Calibration process. Wavelet Neural Network (WNN) brain signal preprocessing algorithm is based on robust algorithms for various noise similar to the ASR algorithm. Since it can be applied to a single channel and its performance is excellent, it is suitable for the real-time system application based on a minimum electrode.
< WNN Algorithm >
< WNN performance>
A study for improving BCI performance - Study on motion image BCI training technique
Motor Imagery imagines movement without physical movement. Relevant areas of the brain include the primary motor cortex (M1), the supplementary motor area (SMA), and the premotor cortex (PMC).
Kinesthetic vs. visual motor imagery
KMI -imagine movement with the feeling that this produces
VMI – imagine seeing the movement
Kinesthetic motion imagery is usually more difficult. However, the accuracy of BCI is higher in kinesthetic motion imagination than in visual motion imagination.
.png)
< Motor Imagery >
ERD/ERS
ERD: event-related desynchronization
ERS: event-related synchronization
%201.png)
%201.png)
< ERD / ERS >
< ERD / ERS - Topography >
BCI technology based on motion image requires high BCI command classification accuracy.
Efforts to improve accuracy are being carried out in various parts.
In addition to brain signal preprocessing, in this laboratory, we are conducting research in various directions, such as a neurofeedback system that helps with motion imagination, and a study on improving accuracy through sensorymotor imagery, so that high results can be obtained by performing motion imagery itself well.
Research for BCI Performance Improvement - Development of Motor Imagery Classification Algorithm
Classification accuracy algorithm improved from 58.0% to 64.7% based on the BCI competition 4-2a dataset (4 types of MI Class), an MI public dataset
Controls remote robots and exoskeleton robots by classifying three types of MI (left, right, foot) in real time based on the developed algorithm
Research to improve BCI performance - Adaptive BCI, Hybrid BCI, Error response
Adaptive BCI is a technology that improves accuracy by performing BCI training based on the user's concentration.
Hybrid BCI is a technology that uses not only brain waves but also eyeball conduction and electromyography, so that the system can be operated and used only with brain waves and biosignals, or it can be used intuitively depending on the application.
Since the accuracy of BCI is not 100%, a technique for responding to errors that may occur
In this laboratory, the accuracy of BCI technology was improved based on the above technologies, and BCI Application was constructed with the goal of actual use.
Application Technology I - Remote Robot Control
Applied Technology Il - Exoskeleton Robot Control
< 2019 Korea Electronics Exhibition Innovation Award >
Applied Technology Ill - Controlling the meal assistant robot
