Control of a Soft Robotic Glove Using a Brain Computer Interface based on Movement Related Cortical Potentials
Author
Frego, Giorgio
Term
4. semester
Education
Publication year
2021
Submitted on
2021-06-03
Pages
12
Abstract
This thesis explores a brain–computer interface (BCI), a technology that allows users to control devices using their brain activity. We tested a Fast Brain Switch (FBS) paradigm designed for quick, binary on/off control. The endogenous sensory discrimination task was supported through a graphical user interface (GUI). The system detected movement-related cortical potentials (MRCPs)—brief brain signals that occur around the onset of movement—using a single-channel classifier, and used these detections as the brain switch. We applied the system to control a robotic glove. In online tests with six participants, the classifier achieved an average precision of 61% (the share of detections that were correct) and an average recall of 68% (the share of intended events that were detected). We also examined how similar the MRCPs were across different movements and introduced a metric to quantify this similarity, which could help with feature selection.
Specialet undersøger en hjerne-computer-grænseflade (BCI), en teknologi der lader brugere styre enheder med deres hjernesignaler. Vi afprøver et Fast Brain Switch (FBS)-paradigme, udviklet til hurtig, binær tænd/sluk-kontrol. Den endogene sensoriske skelnen blev understøttet via en grafisk brugerflade (GUI). Systemet detekterede bevægelsesrelaterede kortikale potentialer (MRCP’er)—korte hjernesignaler omkring bevægelsens start—med en enkeltkanals klassifikator og brugte disse detektioner som hjerneswitch. Systemet blev brugt til at styre en robothandske. I onlinetests med seks deltagere opnåede klassifikatoren i gennemsnit 61% præcision (andelen af detektioner, der var korrekte) og 68% recall (andelen af tilsigtede hændelser, der blev fundet). Vi undersøgte desuden, hvor ens MRCP’erne var på tværs af bevægelser, og introducerede en metrik til at kvantificere denne lighed, som kan hjælpe med udvælgelse af signaltræk.
[This apstract has been rewritten with the help of AI based on the project's original abstract]
Keywords