IEEE Robotics & Automation Magazine - March 2016 - 92

20
0

20
0

- 20

-20
-40

Right Hip Angles
(°)

Left Hip Angles
(°)

0

20

40

60

80

100

- 40

0

20

0

20

0

20

40
60
80
Gait Cycle (%)
(c)

100

0

20

40

60

80

0
- 20
- 40
- 60
- 80

115
110
(% TM)

110
105
100

(a)

100

Gait Cycle (%)
(e)

0

20

(b)

Figure 7. The (a) walking speed and (b) cadence mean values
and standard deviations of the overall group for each of the
trials, normalized to the TM trial.

the case of LSA, there was no significant reduction either in
speed nor in cadence (p = 0.1823 and p = 0.0710, respectively). In any case, when comparing the TM mode with any
of the assistive trials, there was no significant difference.
Wearing the device caused the walking pace to decrease,
and no change was observed between the nonassisted and
assisted trials.
Interestingly, this connects to a nice model-based prediction: if we assume that the subjects walk at their resonance
frequency (i.e., the one that
optimizes the exchanges between kinetic and potential
NO
TM
energies), increasing their
LTA
mass when wearing the APO
HTA
LSA
should make this frequency
HSA
[23]
decrease. Indeed, for a massspring system (which is often
used as a very simplified
40
60
80
100
model of walking [32]), the
Gait Cycle (%)
resonance frequency is pro(b)
portional to 1/m . Therefore, it is logical to observe a
decrease in walking cadence
when the body mass is augmented, i.e., when the subject
wears the device.
40
60
80
100

40

60

80

100

Gait Cycle (%)
(f)

Figure 6. The kinematics of all joints during the 2 min of steady-state walking: (a) left hip, (b)
right hip, (c) left knee, (d) right knee, (e) left ankle, and (f) right ankle. The colored solid lines
represent the different trials the subjects underwent. The dashed black lines display data from
[23] from the literature for comparison. Positive values represent extension, and negative values
represent flexion. This figure shows data averaged across the healthy participants.

92

*

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

march 2016

N
O
TM
LT
A
H
TA
LS
A
H
SA

90

N
O
TM
LT
A
H
TA
LS
A
H
SA

95

90

40
20

0
- 20
- 40

105
100

95

Gait Cycle (%)
(d)

40
20
0
-20
-40

Right Knee Angles
(°)

0
-20
-40
-60
-80

Right Ankle Angles
(°)

Left Ankle Angles
(°)

Left Knee Angles
(°)

Gait Cycle (%)
(a)

115
(% TM)

participant. The walking cadence was estimated from the frequency continuously learned by the AOs from the hip angular position. The average subject's cadence during the 2-min
walking interval was thus computed as the median value of
the average of the right and left legs' frequency over time.
We observed that wearing the device (with or without assistance) tended to make the subject walk more slowly than
in the NO trial. This was correlated with a similar decrease
in walking cadence. In general, the overall walking speed
can be decreased (respectively increased) in two concurrent
ways: either by decreasing (respectively increasing) the
walking frequency or by decreasing (respectively increasing)
the step length. It seems that our participants favored the
former, although no direct measurement of the step length
was available. Statistically significant differences for the
walking speed with the NO trial were reached for the LTA
and HTA trials ( p = 0.0313 and p = 0.0137, respectively).
The same trials displayed a significant decrease in the walking cadence (p = 0.0113 and p = 0.0084, respectively). In
sum, there was a significant reduction of both walking speed
and cadence in the LTA and HTA trials. In the case of HSA,
there was a significant reduction of the walking cadence
with respect to the NO trial (p = 0.0.0152); however, the
speed reduction did not reach significance (p = 0.0533). In

Walking Dynamics
The average joint reference
torques computed in the different trials are shown in Figure 8. In the case of NO, the
APO was not worn by the
subject, and, thus, no torque
was provided. In the case of
TM, the torque reference was
set to zero, meaning that the
powered orthosis was asked
to follow the intended movement of the person without
hindering it, i.e., with no
Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2016
