I collaborate with the BCI group at the University of Zaragoza on the development of adaptable BCI systems through the use of learning algorithms. I am involved mainly in three different research topics:
Learning from brain signals
Error related potentials occur spontaneously in the brain while observing or performing a task. They supposedly play a role in human learning as implicit feedback signals that evaluate the correctness or unexpectedness of received stimuli. Our goal is to reliably detecti and classify these signals to provide automatically feedback to artificial systems (e.g. a robot) that learn how to interact and adapt themselves to the user intentions and preferences.
Undestanding and decoding human motion
There is still an ongoing debate whether it is possible to decode motion from EEG measurements. Although recent works have shown that it is possible to correlate EEG to reaching motions, for instance, we are still far away from robustly decoding motion from EEG data in a way that can be used within other systems. Potential applications include rehabilitation, robot control, video games and any kind of tele-presence.
We are currently working on designing good features for regression that allow to reproduce training trajectories, but also allow to generalize to unseen ones. The video on the right show the simultaneous recording of the trajectories and EEG data.
BCI for Rehabilitation
BCI systems have promised for a long time to play a key role in rehabilitation. In the HYPER project, we are studying how BCI information can help therapists to improve the rehabilitation procedures for stroke patients and for SCI (Spinal Cord Injuries). We are currently investigating how to create brain activity in this type of patients that can be used in the rehabilitation sessions.