Page 148 - Handbook of Biomechatronics

Biomechatronic Applications of Brain-Computer Interfaces 145

could use a BCI to control the wheelchair only with their mind, thus mov-
ing around with any assistance from a caretaker.
Depending on how much authority is left to the users, several wheelchair
BCI architectures can be considered. For example, one P300-based BCI
wheelchair includes a screen (mounted on the front or side of the wheel-
chair) that displays a 3 3 grid of possible destinations in the user’s house
(e.g., the bathroom) (Rebsamen et al., 2010). The rows and columns are
sequentially highlighted, and the desired destination triggers a P300
response. Once the BCI has identified the desired destination, the wheel-
chair autonomously moves to that room along a predefined route, though
the user can send a mental “emergency stop” command to terminate the
movement. This greatly simplifies the BCI functioning, but limits the user
to a few predefined locations that they can access.
Wheelchairs with more autonomy allow the user to perform individual
commands such as “move forward,” “turn left,” etc. This can be done with
several different BCI paradigms. For example, a common strategy for wheel-
chair control is via SSVEPs induced by a screen mounted on the front or side
of the wheelchair. Several buttons labeled “move forward,” “turn left,” etc.
are presented to the user on the screen, with each button flashing at a dif-
ferent frequency. The user selects the desired command by gazing at the
corresponding button, causing an SSVEP of the same frequency in the
occipital lobe, which is detected by the BCI or sent to the wheelchair.
The wheelchair then has different options regarding how to respond:
• it can carry out one discrete command (e.g., move 3feet forward), stop,
and wait for the next one,
• or it can keep executing the command until the user either stops looking
at the screen (resulting in no SSVEP observed the BCI) or looks at a dif-
ferent button on the screen.
Both approaches have their own advantages and disadvantages. If the wheel-
chair executes discrete commands, it tends to be stationary much of the time
while waiting for the next command. Conversely, if the wheelchair keeps
executing the command until the user changes their gaze point, there is
higher potential for accidents, for example, the user may keep looking at
the screen and not realize that the wheelchair is about to hit an obstacle.
An advanced approach that utilizes motor imagery and aims to reduce
the user’s mental workload was presented by Carlson and Milla ´n (2013).
In brief, the wheelchair responds to two different types of motor imagery
that correspond to turning the wheelchair left or right. However, if neither
type of imagery is detected by the BCI, the wheelchair continues moving