Page 345 - Handbook of Electronic Assistive Technology

334 HANDBOOK OF ELECTRONIC ASSISTIVE TECHNOLOGY

torque that the actuators can apply. Additional sensors can be included to detect the malfunc-
tioning of other sensors and implement a safe protocol. Overall, the system should be fail safe,
meaning it should be able to achieve a safe state in the presence of a detected fault. When a
fault is detected, the exoskeleton should either stop motion and hold the current position, or
remove power to the motors (Roderick and Carignan, 2005; Baniqued et al., 2015).
Safety features that can also be incorporated in the software include current and speed
monitoring, collision detection, routines to interrupt the power supply to the motors, soft-
ware checks to limit forces, motions and speeds and user adjustments to control param-
eters, as well as checks for sensor health and other dangerous situations (Nef et al., 2006;
Van der Loos and Reinkensmeyer, 2008).
It should also be possible for the robot to be moved manually by a therapist to release
the patient from a potentially uncomfortable or dangerous position (Nef et al., 2006). This
can be achieved by using backdrivable hardware (Babaiasl et al., 2015). Achieving both
high-force production capability and backdrivability is an engineering challenge in reha-
bilitation robots.
Biomechanical function: Exoskeletons are wearable devices that operate mechanically
on the human body, with possible interference and friction with limb natural movement
(Bruni et al., 2018). Therefore an understanding of the biomechanics of the joints and the
biomechanics of human walking is essential in the design of exoskeletons for the lower
limbs (Dollar and Herr, 2008; Low, 2011). Although robotic joints are generally designed to
mimic human joint kinematics in terms of structure, range of motion and DOFs, the com-
plex nature of human joints means that robotic joints tend to be simplified. This can lead
to low kinematic compatibility between the human and robotic joint causing unwanted
interaction forces between the human and exoskeleton. However, increasing the complex-
ity of the joint may lead to increased costs and reduce the reliability of the system.
Similarly, a high number of DOFs allow a wider variety of movements, with many ana-
tomical joint axes involved (Nef et al., 2006). However, trade-offs exist between the num-
bers of DOFs to provide the range of requirement movements and the size, weight and cost
of the device. The human leg can be approximated into a structure with a total of seven
DOFs: three rotational DOFs at the hip, three at the ankle and one at the knee (Calabrò
et al., 2016). The upper limb effectively has a total of nine DOFs, excluding the finger joints.
Ideally, the exoskeleton should be kinematically compatible with the human joint while
still providing satisfactory functionality (Low, 2011).
The hip, knee and ankle are weight-bearing joints, which rely on sufficient muscle force
for stability. Mobile medical exoskeletons should therefore provide sufficient external joint
moment to compensate the lack of forces in these joints and also provide BWS to minimise
the weight loaded on these joints (Low, 2011).
The user should not feel the weight of the robot. The robot must be capable of generating
sufficient force to move a patient’s limb, and they should be able to move the device easily.
Autonomy/shared control: A shared control system must have the ability to determine
the user’s intention, verify that the desired action to be performed is safe, and when appro-
priate be able to adjust the control signal to manoeuvre the device safely and efficiently
(Carlson and Demiris, 2012). In essence, this is a division of labour between the user and