A multi-modal BCI system for active and assisted living

Brain Computer Interface (BCI) technology is an alternative/augmentative communication channel, based on the interpretation of the user’s brain activity, who can then interact with the environment without relying on neuromuscular pathways. BCI can thus be placed in the context of human-machine interfaces and, considering a possible application scenario to smart homes, they can serve as a technological bridge to make Active and Assisted Living (AAL) systems’ functionalities accessible to subjects who would not otherwise be able to actively use. In this paper, BCI is specifically conceived for AAL system control experience, developed ad-hoc, considering cost and compactness constraints, besides classification performance. The implemented solution is quite general, as it can handle multiple bio-potentials: this feature allows to exploit different information channels, namely, ElectroEncephaloGraphy (EEG) and ElectroMyoGraphy (EMG). Each subsystem (EEG, EMG) is presented and its performance discussed; both can operate in real-time and in self-paced mode (i.e. they automatically recognize if a command is being issued and, in this case, which one). In particular, the EEG part, based on Steady State Visual-Evoked Potentials (SSVEP), can achieve very good results in terms of false positive rejection, improving over the state of the art (False Positive Rate: 0.16 min−1). Moreover, such results are achieved without any initial system calibration phase. Meanwhile, the EMG subsystem can be used as a smart switch for SSVEP stimuli control (on/off), improving user’s comfort when no control periods are desired