US 11,914,773 B2
Brain-machine interface based intention determination device and method using virtual environment
Seong-Whan Lee, Seoul (KR); Byoung-Hee Kwon, Seoul (KR); Ji-Hoon Jeong, Seoul (KR); Kyung-Hwan Shim, Seoul (KR); and Byeong-Hoo Lee, Seoul (KR)
Assigned to Korea University Research and Business Foundation, Seoul (KR)
Filed by Korea University Research and Business Foundation, Seoul (KR)
Filed on Jan. 20, 2021, as Appl. No. 17/152,909.
Claims priority of application No. 10-2020-0019493 (KR), filed on Feb. 18, 2020.
Prior Publication US 2021/0255706 A1, Aug. 19, 2021
Int. Cl. G06F 3/01 (2006.01); G06N 20/00 (2019.01)
CPC G06F 3/015 (2013.01) [G06N 20/00 (2019.01)] 16 Claims
OG exemplary drawing
 
1. A device configured to perform an intention determination process, the process comprising:
receiving an indication of a thought mode selected by a user input;
selecting content according to the indication of the thought mode;
displaying the content while collecting a first brain signal sensed from the user while the user is viewing the displayed content;
extracting a first filtered signal from the first brain signal using a frequency band filter selected based on the indication of the thought mode, wherein the thought mode is selected from predefined thought modes, and the frequency band filter is selected from a plurality of frequency band filters each associated with the predefined thought modes;
generating physical information representing a physical magnitude of the extracted first filtered signal by calculating a first power spectral density from the extracted first filtered signal;
determining control information according to the physical information based on a classification model generated by learning control information representing a user's intent according to a pattern in the physical information provided in advance;
outputting a control command when the control command matches the control information; and
when the control command does not match the control information, regenerating the physical information by calculating a second power spectral density based on a second brain signal having a time and frequency different from the first brain signal,
wherein the regenerated physical information is input to the classification model to generate matched control information.