BCI : An interface designed to control a technical device by translating brain activity into commands without a direct dependency on or use of any muscular activity.
- A new interface for HCI (Human computer interaction)
Modes of interaction in BCI Active & Reactive BCI : Direct Control example :Using P300 to control keyboard Passive BCI : Implicit Control
user modeling with BCI
Neuroadaptivity :
Generating a BCI
-
User trained BCI
- The o/p of the the machine defines the i/p to the BCI. It expects a specific command to be sent for executing the related action.
- The user needs to train how to generate the pre defined brain activity at the sensor. Once user learned to control their brain activity precisely a stable communication through the BCI is possible.
-
ML based BCI
- Diff i/p commands are generated by the user and the BCI learns to detect the difference between them.
Measuring brain activity Some approaches include fMRI, fNIRS, MEG, EEG MEG : not used much, needed to be cooled Classic neuroscience for interpreting brain activity : Isolating process, avg over population as brains are highly variable among a population.
With BCIs there is no isolation, no averages so there is more noise. The goal is more to predict than explain as in classic neuroscience.
Applications of BCI
- Passive BCIs
- Mental State Assessment :
- Workload monitoring at the office
- Mental State Assessment :
- Closed loop : system adapts
- Open loop : User gets a feedback but system does not adapt.
Motor Imagery Control (Active BCI)
- C3, C4 used from EEG
- Classification done with LDA