Projects

This page briefly introduces some of the projects that Team PhyPA has worked on, or is currently working on.

BRAINFLIGHT: Controlling a flight simulator by thought (2014)

One investigation into the use of brain-computer interfaces in the real world was done as a part of EU-funded project BRAINFLIGHT. An active motor imagery BCI was used to control the the fly path of an aeroplane inside a simulator: By imagining different kinds of movements, pilots could steer the aeroplane either to the left or to the right. Altitude and throttle were automated.

Seven professional pilots and pilots in training conducted both an instrumental flight (looking only at their instruments) and a visual flight (looking outside). A number of different flight exercises taken from official pilot training programmes were used. The final task was a critical landing procedure that required large adjustments of the flight path on short notice.

Some participants using the BCI did well enough, using official performance measurements, to have passed if it were an exam. These results appear to support Team PhyPA's hypothesis that such realistic conditions, due to heightened participant motivation and involvement, increase BCI performance. This supports Team PhyPA's goal to meaningfully apply BCI technology outside of the laboratory, in everyday tasks.

Related links and project partners:

In-car EEG and BCI (2008 - 2014)

Applying EEG and BCI technology outside of the laboratory means having to deal with movement artefacts and noise—certainly so in a fast-moving car on imperfect terrain. However, passive BCI may be able to significantly improve a driver's experience and safety, for example by monitoring the driver for signs of drowsiness. Team PhyPA has been investigating the pitfalls and opportunities of such scenarios for a number of years.

Related links and project partners:

Tracking taskload of surgeons (2013)

A useful application of passive BCI in a real world scenario is the measuring of cognitive load of a surgeon in the theatre. Cognitive load is a covert aspect of the surgeon's state that can be detected automatically, without any additional action on the side of the surgeon. Knowledge of the surgeon's current cognitive load could then be used to automatically support the surgeon during stressful actions, and to communicate to the assistants when the surgeon is engaged in a complex maneuver and should not be disturbed.

Skilled surgeons performed simple and complex surgical training tasks. Results indicate that a passive BCI can reliably and continuously detect changes in cognitive load in this realistic environment.

Related links and project partners:

Error recognition and correction: The RLR Game (2012)

Human-computer interaction may be significantly improved when the errors that are committed are automatically detected and corrected. Indeed, a passive BCI may be able to automatically detect in the user's EEG whether or not an error has just occurred. With this information, the error can then be corrected by the system, without requiring any manual corrective action from the user.

Participants—tens of them, over the years—played a game in which they rotated letters clockwise or counter-clockwise in steps of different size until the letters were in their correct positions. However, the game was programmed to sometimes rotate the letters incorrectly: The system would make mistakes. A passive BCI was calibrated to detect these errors and, when detected, correct the mistakes in the rotation.

The results show that a passive BCI can meaningfully support human-computer interaction by detecting and correcting errors.

Related links and project partners:

Combining active and passive BCIs with eye tracking (2009)

Using eye tracking technology, the eyes can be used as a natural replacement for classic pointing devices such as the mouse. However, there is no natural gaze-based command that represents the selection of what is being pointed (or looked) at. Both active and passive BCI may offer solutions.

Using an active BCI, a specific selection command could be trained and used. For example, when the user imagines a grabbing movement, this can be interpreted as a selection.

Passive BCIs, alternatively, can automatically detect the intention to select, without requiring any specific additional commands from the user.

Related links and project partners:

Classic motor imagery: Basket paradigm (2006)

The classic example of an active BCI involves the horizontal control of a cursor. When imagining a right-hand movement, the cursor is moved to the right; when imagining a left-hand movement, the cursor is moved to the left. Using Team PhyPA's basket paradigm, participants were able to so control a cursor with high accuracy.

Related links and project partners: