Collaborative robotic biomechanical interactions and gait adjustments in young, non-impaired individuals

OUT vs AM conditions

The first aim of this study was to compare the joint kinematics, forces and moments
during walking at a fixed constant treadmill belt speed and constrained walking cadence,
with and without the robotic device. We hypothesized that the relative transparency
of the robotic system would allow for similar kinematic trajectories, without interruption
and/or direction reversals, however, due to external forces exerted by the device
on the user in the forward propulsive direction, we expected changes to ground reaction
forces, and joint moments “in” versus “out” of the device during assistive mode (OUT
and AM conditions). The results partially supported our hypothesis, since the phasic
nature of the kinematic characteristics of walking (i. e. support and swing phases)
was similar when comparing OUT and AM conditions in sagittal and frontal planes (Figs. 2a, c and 4a, c). These results confirmed the observations with functional activities in this device
12], 15]. However, in the AM condition there was more angular displacement for the ankle and
hip joints in sagittal plane. There were also important kinematic magnitude differences
during mid-stance in the sagittal plane. In the AM condition, the trajectory of the
pelvic mechanism was in an inferior direction after heel strike and we found the lowest
value in mid-stance, about an average of 0.04 m (Fig. 3c). However, the vertical force indicated an upward direction (Fig. 3b). This is because the belts are attached to the participant, and the lower displacement
of the pelvic mechanism produced tension in the belt indicating a force amplitude
direction in the upper direction. So, the actuated movement of the pelvic mechanism
in the vertical direction appears to have pushed the individuals downward against
the treadmill, increasing the vertical force (maximum values in mid-stance), and affecting
the vertical impulse (Table 3; Fig. 3a). The pelvic mechanism also purposefully limited the anterior displacement of the
body (so that interaction forces can be picked up the force sensor in an isometric
position), generating a force in posterior direction (Fig. 3f) between early stance and mid-stance sub-phases. The fixed position of the pelvis
in the anterior-posterior direction imposed reduction of the braking force, which
was reflected in reduced braking impulse (Table 3; Fig. 3e). On the other hand, the flexor and extensor hip moments were increased, presumably
to provide stability and acceleration of center of mass in the forward direction.
Another interesting aspect is that the knee moment in both conditions was flexor during
all of the stance phase (Fig. 2e). The possible explanation is that, over the treadmill, the braking force was slight
reduced, especially with the slower speed used in this study (below the comfortable
speed) 16]. The reduced braking force and reduced knee moment during walking on the treadmill
at a comfortable speed was also observed in a previous study 17]. In the frontal plane, during the AM condition, maximum values for lateral force
and hip moment were reduced (Fig. 4e). Taking everything together, the main interactions occurred during mid-stance, where
braking force was reduced, but vertical force and hip extensor moment were increased.
The movement of the pelvic mechanism in the vertical direction and the reaction force
from the pelvic harness system reduced the braking and lateral forces. As a result,
the harness system reduced the center of mass displacement (Fig. 5a). Our results are consistent with a previous study, where we observed a reduction
of the braking force by horizontal force that was applied close to center of mass,
and it was proportional to the resistance that was applied 18]. The participants’ response to forces from the pelvic mechanism served to increase
the extensor and flexor hip moments. The simple instruction to the participant to
increase the push off provoked the increased hip moment and angular displacement 19]. In this study, the resistance imposed by pelvic mechanism also effectively generated
larger hip moment and angular displacement.

AM vs SDM conditions

The second aim of this study was compare AM and SDM conditions in device and determine
the relative impact of user-generated forces, sensed by the pelvic mechanism, that
drive the treadmill. We hypothesized that, when the person attempted to control a
target speed, they would need to adjust kinematics and kinetics in response to interacting
with the pelvic mechanism in order to perform the task compared to the AM condition.
The results confirmed our hypothesis, since the results for the sagittal plane showed
that, in the SDM condition ankle and hip joints had larger angular excursions and
the knee showed larger flexion and smaller extension (Fig. 2a, c; Table 2), which represented a more flexed posture in the SDM condition. Vertical force (Fig. 3a) was not affected, but there was reduced braking force and increased propulsive force
in the SDM condition (Fig. 3e). Similarly, the extensor hip moment was increased for the SDM condition compared
to the AM condition (Fig. 2f). This situation was reflected by impulse measures, which were reduced for braking
impulse, but larger for propulsive impulse (Table 3; Fig. 3e). For the frontal plane, the results showed that larger hip angular excursions appeared
with increased hip abduction in the SDM condition (Fig. 4c).

The main results for this comparison occurred during mid-stance and late stance. The
challenge imposed by the SDM condition lead to a more flexed posture compared with
the AM condition. In addition, the participants had to apply the appropriate force
over the sensor embedded in harness system, to drive the treadmill belts at a particular
speed. The pelvic mechanism also generated a force in the posterior direction (Fig. 3f) between early stance and mid-stance sub-phases and the inferior trajectory of the
mechanism (Fig. 3d), showed a reduced braking force 18], increased extensor hip moment 19] during mid-stance (Fig. 2f) and propulsive force during late stance (Fig. 2c). The propulsive force increase was observed in a previous study and tended to be
proportional to the horizontal resistance applied during walking 18]. So, the participants of this study in SDM condition had additional adjustments compared
to AM condition.

Overall human-machine interaction for collaborative robots

While the goal of a robotic walking device is to assist with movement without interfering
with the basic mechanics of walking, the nature of the human-machine interface is
very difficult to overcome in practice. There are a number of recently developed robot
devices applied to improve gait, and in general, all devices have limitations like
movements constrained to one anatomical plane (sagittal) which prevents meaningful
balance training, reduced degrees freedom on pelvis and or trunk, where the patient
is guided during movement 6]. Even with guided movement, non-impaired individuals showed altered angular displacement
in lower limb 20], 21] and trunk 22] tested with and without others gait orthosis devices, sometimes increasing 20] and sometimes reducing 21], 22] the movement. In the case of the KineAssist applied over a treadmill, the basic phasic
trajectory of the kinetic and kinematic variables were intact (i. e. natural walking)
and without arresting or reversing movement trajectories, however important interactive
forces were imposed. Despite these interactions, the center of mass displacement had
a consistent trajectory (Fig. 5b) and participants were able to respond to interactive forces with mostly small adjustments.
Perhaps, these adjustments could serve to increase the rate of learning, at least
temporarily, during the execution of a task 23].

The results from this study highlight some important advantages and disadvantages
associated with a collaborative robotic system such as the KineAssist. In terms of
advantages, the system allows the user to drive the movement of the treadmill belt
in SDM which encourages active engagement in the task. Also, while not studied here,
the device allows safety and confidence for people with poor balance who are regaining
walking ability, since the device will catch a person when their pelvis height drops
below a set height. With AM, individuals with reduced force generation during propulsion
are afforded opportunities to walk at faster speeds. In terms of disadvantages, the
device will reduce comfortable walking speed by as much as 50 % 15] and will require greater mechanical work to raise the vertical trajectory of the
center of mass. These two issues will cause extra fatigue and potential muscle soreness
when a person exercises with the device. The most obvious disadvantage comes from
the caution that any researcher and/or clinician must use when interpreting the trajectory
characteristics of the gait pattern as the results of walking in this device should
not be used to diagnosis specific gait pattern deficits in individuals with impaired
walking.

Limitations

There are some limitations to be considered in this study. First, the experiment was
conducted at one fixed speed which may or may not reflect the person’s comfortable
walking speed. However, the speed of 1.0 m/s that we used in this experiment is slower,
but close to reported average comfortable walking speeds 16]. Second, we tested at certain damping and deadband settings, and the results could
be different with alternatives levels of damping and deadband. A lesser deadband level
may have resulted in less effort to drive the treadmill during the SDM condition,
while lesser damping settings may have provided greater velocity-dependent sensitivity.
However, we selected the particular settings used in this study in order to permit
the optimal control of treadmill steadiness and stability. Third, the device used
in this study was an older prototype version of the system and has a more massive
pelvic mechanism than more current systems that are now on the market. Finally, the
participants of this study were healthy and young, classified with moderate to high
level of physical activity (Table 1), which provided ideal candidates for adjusting their gait characteristics in the
device. Adjustments observed here could be modified in different impaired or older
populations but these hypotheses need to be tested with future investigations.