Page 320 - Handbook of Biomechatronics

312 Andres F. Ruiz-Olaya et al.

simultaneous multifunctional control that would be widely accepted by
users (Ison and Artemiadis, 2014). Pattern recognition-based schemes com-
monly require an initial adaptation to the outputs when controlling the
device in real time. Gibson et al demonstrated that a control scheme trained
on a variety of users can extract the low-level population-wide synergies and
provide good performance in offline analysis, and better performance in
real-time given visual feedback (Ison and Artemiadis, 2014). Recent
implementations of simple control schemes based on extracted synergies
have shown robust performance compared to more complex classifiers.
These results suggest that intuitive, user-independent control schemes can
be developed to provide user-friendly, low-level control without requiring
an intense training phase from the user (Ison and Artemiadis, 2014).
Similarly, recent trends and attempts in developing electroencephalog-
raphy (EEG)-based control methods have shown the potential of this area
in the modern bio-robotics field. A new approach of combining both con-
trol methods, which use the advantages, and diminish the disadvantages, of
each system might therefore be a promising approach (Lalitharatne et al.,
2013). In this case, EEG signals can be used to compensate for insufficient
information in the EMG signals. Numerous examples such as wheelchairs,
prosthetics, exoskeletons/orthoses (Lalitharatne et al., 2013) show the
effectiveness of EMG-based control methods. However, these EMG-based
control approaches used alone have some disadvantages that depend on the
user and on the application. In cases where the user cannot generate suffi-
cient muscle signals, EMG-based methods are not useful for movement
intention detection. For example, a person who has a totally paralyzed upper
limb may not be able to use a device such as an exoskeleton due to the dif-
ficulty of getting control signals from the muscles of the paralyzed limb. In
this case also, EEG can be used to compensate for the missing EMG signals
(Lalitharatne et al., 2013). Even if all required muscles for EMG are available,
EEG can still be used to remove the effect of fatigue or undesired tremor.
Applications of the hybrid approaches may vary from a simple game control
application for an able-bodied person through to a prosthetic arm and
exoskeleton control application for an amputee or motor disabilities person.
Technology is one of the limiting factors for hybrid EEG-EMG-based
control approaches. High-density EEG systems can provide a lot of details,
but it is sometimes not practical to use such systems when they cover the
whole head of the user, as the user may feel uncomfortable (Lalitharatne
et al., 2013). Compact and low-weight designs for EEG and EMG data mea-
suring systems need to be introduced, in order to allow use when users need