Page 280 - Control Theory in Biomedical Engineering

252 Control theory in biomedical engineering

upper arm internal/external rotation joint is at 0 degrees and forearm pro-
nation/supination is at 0 degrees. The human upper limb has natural singu-
larity, but it does not create difficulty in moving the limb away from a
singular position. However, unlike the human upper limb, actuators in an
exoskeleton require infinite torque to move itself from a singular position.
Some researchers did not consider this issue because it is rare to encounter a
singular position in a rehabilitation protocol (Perry et al., 2007; Carignan
et al., 2007). Nonetheless, the exoskeleton should be rid of singularity,
otherwise it can be stuck in a singular position. There are two areas where
effort can be given to address singularity. Researchers might design an exo-
skeleton’s structure so that it evades singular position, or a control algorithm
might be developed as to avoid singular position in exoskeletons.

Backdrivability
For therapy where patients remain passive, actuators in an exoskeleton does
not need to be backdrivable. In contrast, therapy where patients need to
participate actively (i.e., patient contributes to the movement of the exoskel-
eton) in control sharing, actuators must be backdrivable. With intensive
rehabilitation therapy, patients start regaining lost mobility in upper limbs.
Hence, it is possible to engage them actively in the rehabilitation process.
Therefore, upper limb exoskeletons must allow patients, when they are able,
to move their limb on their own during therapy sessions (Garrec et al., 2008;
Sutapun and Sangveraphunsiri, 2015).

Sensors

Sensors plays a big role in controlling exoskeletons. Incorrect sensor values
may hurt the patient or even lead to injuries. Most of the existing exoskel-
etons use joint position and joint torque in their control approach. To get
the position, IMU, potentiometer, and hall sensor were used in the exoskel-
eton system. Six-axis force sensor was used at the wrist to asses both the force
and torque exerted by the user onto the exoskeleton. The 3-axis force sensor
also can be used to get the Cartesian force at wrist; in this case Jacobian of the
exoskeleton can be used to find the joint torques. This control technique
requires more computation as it has to compute the system’s Jacobian.
In addition, EMG-based control uses muscle signals to detect the user’s
movement intention (Liu et al., 2020; Xiao, 2019; Priyadarshini et al.,
2018; Accogli et al., 2017; Li et al., 2017; Peternel et al., 2016; Gopura
et al., 2009). A noninvasive surface EMG electrode was used to obtain