< Research activities \>

Videos with applications of brain-computer interface (BCI)

Videos of some experiments with different BCIs (communication spellers and control of a robotic wheelchair). See publications for details.

Lateral single-character speller (LSC) communication speller

LSC is a P300-based BCI that minimizes some negative effects of the standard row-column speller such as distractors, eyestrain, high target probability (see publications). Moreover, thanks to the events strategy and layout, there are new neurophysiologic features with discrimination power. The video shows an able-bodied participant spelling the portuguese sentence "Bom dia a todos" around 10 symbols per minute. This BCI has been tested by participants with motor disabilities, namely, amyotrophic lateral sclerosis, cerebral palsy, Duchenne muscular distrophy and spinal cord injury.

Gaze Independent Block Speller (GIBS) communication speller

GIBS is a P300-based brain-computer interface (BCI) that allows to select letters with minimal (or none) gaze shifts. GIBS requires a 2-level selection. The central letters are in the fovea region, therefore user perceives the target without gazing. Small blocks are outside the fovea region, but user only needs to know in which block is the letter. Thus, the percepetion of the target-block stimulus is independent of gaze. EOG is being monitored to detect eye movements.

Row-column communication speller

Video of two participants with motor disabilities (cerebral palsy and Duchenne muscular distrophy) controlling the row-column speller. The sentences are in portuguese, respectively, "Parabens APCC" and "Estou a escrever".

Brain-actuated wheelchair (RobChair 1.0)

Brain-actuated wheelchair using a P300-based BCI. Pilot experiments with an able-bodied user in the wheelchair and a participant with cerebral palsy controlling remotely the wheelchair (due to ergonomic and confort constraints). Shared-control system: the wheelchair navigates in pre-defined trajectories and accepts brain commands only in decision points (e.g., bifurcations, dynamic obstacles) . The inputs from the user are steering directions such as "left", "forward", "right", etc. (see the full video in robchair.isr.uc.pt )

Brain-actuated wheelchair (RobChair 2.0)

Collaborative control system: RobChair is controlled with a P300-based BCI in a self-paced mode (participant only conveys commands when he/she wants to). Navigation system performs manoeuvres in narrow spaces. The inputs from the user are steering directions such as "left", "forward", "right", etc. 



Ongoing research

- Design and implementation of new P300 visual paradigms searching new neurophysiological features
- Design and implementation of gaze independent P300 visual paradigms
- Self-paced control operation
- One-time calibration / zero time calibration
- Development of signal processing methods for feature extraction based on statistical spatial filters and spatio-spectral filters
- Multimodal interfaces: Fusion of EEG, EOG, EMG and inertial sensing
- Applications: communiction spellers, icon selection, control of a robotic wheelchair, game control
- Evaluation of BCI by motor disabled end-users (cerebral palsy, amyotrophic lateral sclerosis, Duchenne muscular distrophy, spinal cord injury). This research topic counts on collaborations with the Cerebral Palsy Association of Coimbra and with the Hospitals of the University of Coimbra



Other Assitive interfaces

I have worked for some years in the developement of shared control algorithms for robotic wheelchair (RobChair) steering. The focus was on the development of reactive controllers to perform obstacle avoidance manoeuvres while receiving steering commands from the user through voice (see some videos below and at the RobChair web page).


RobChair steered by voice commands A reactive fuzzy logic controller receives voice commands from the user and combines this input with environment perception to perform obstacle avoidance.