Analysis of ankle kinetics and energy consumption
with an advanced microprocessor controlled
ankle­foot prosthesis.
D.Moser, N.Stech, J.McCarthy, G.Harris, S.Zahedi, A.McDougall
Summary
This study reports on several experimental investigations carried out to analyse the impact of
adaptive control on the kinetic behaviour of ankle foot prosthesis in conjunction with locomotion
energetics and biomechanical performance. The healthy limb has an amazing capability to adapt to
the changing requirements of walking such as walking on varied inclined ground and at different
walking speeds. Recently advances in prosthetics have sought to design prosthetic feet that have
some form of adaptive capability combined with a deeper understanding of the effect of viscoelastic
(hydraulic damping + elastic) foot behaviour. It is already now established that this functional
concept provides greater stability within the socket, improved proprioception and lower stump
interface pressures (Portnoy 2011). Crucially it has also been shown that more balanced distribution
of loading takes place between pathological and intact limb thus reducing the risk of further
pathologies developing and reducing healthcare costs. It is clear from biomechanics research that
the “rolling­collision” between foot and ground during locomotion is of paramount importance as
biomechanical inefficiencies created at this interface are magnified and propagated up through the
limb which may require greater compensatory effort on the part of the amputee. It seems logical
that the changing environment and subtle changes between steps introduce a performance variable
which must be properly understood if technology is to advance further. For further biomechanical
optimization it seems logical that the mating foot in this collision must adapt in some way to the
changing requirements to provide greater biomechanical optimization.
Methods
An advanced microprocessor controlled ankle foot system called elan (see figure 1) has been
developed that has the capability to automatically alter and adjust how the foot interacts with the
ground by controlling the amount of energy that is stored and released elastically by the foot in the
stance phase of the gait cycle. The novel paradigm explored, is that active manipulation of the
“rolling­collision” that takes place between foot and ground can be used to optimize biomechanical
performance for walking on various terrains and activities of daily living (ADL). In order to alter the
characteristics and exchanges of energy between the body and ground changes in viscoelastic
properties can be used to create more or less elastic effects and biasing effects to either “assist” or
“brake” the fore­aft exchange of energy as the body mass transitions over the ankle. The assist
effect increases the hydraulic damping resistance in the PF direction whilst simultaneously the DF
damping resistance is reduced. The net effect is that the heel spring becomes “more elastic” and the
damping resistance in the DF is reduced to enable the greatest possible release of energy and
minimize any hindrance at the ankle for anterior progression. The aim is to introduce this assisting
effect advantageously when walking at faster speeds and up inclines to make the walking task less
demanding. The brake effect reduces the PF damping resistance and increased the DF damping
resistance, the net effect is that the heel spring becomes less elastic and by increasing the DF
resistance, motion and energy transfer in the anterior direction is “braked”. The aim is to introduce
this braking effect advantageously when walking down inclines to create greater stability and safety
and to reduce the demand on the residual musculature to provide the same braking effect. The
design aim is that although only one set of carbon fibre spring is used, by adapting the hydraulics
and this energy transmission through the springs “virtual spring ratings” can be created, in other
words the “springy” effect of the springs can be adjust producing the same response as if the springs
themselves were replaced with different spring constants. These virtual spring ratings change during
the swing phase of the gait cycle so that at the point of heel strike the complete visco­elastic system
is fully optimized.
Figure 1, Elan, microprocessor ankle­foot system
The experimental emphasis of the studies reported here were aimed towards developing a deeper
understanding of how changing ankle foot properties via assistive and braking effects automatically
for different walking environment and gait speeds influences the kinetics at the ankle and efficiency
of locomotion produced by amputees. Pylon load cell, GRF were used to quantify ankle kinetics, and
on board joint angle sensing data was user evaluated damped motion characteristics. High speed
digital photography was used to quantify changes in spring defections. For each walking
scenario/task the ankle foot properties (damping settings) were altered in each case to determine
how the ankle foot kinetics/kinematics would change. Tests were contacted at various damping bias
settings (assist) and (brake) and neutral (balanced PF/DF resistances) A wireless data acquisition
system was used to collect the data and Matlab was used for data post process and statistical
analysis of the results.
Results and discussion
Independent Research carried out at the Gait Analysis Lab of the University of Surrey, explored the
improvements in the efficiency of walking when prosthetic feet with the introduction of visco­elastic
ankle­movement in the sagittal plane were compared to the conventional, purely­elastic type
(Khadra, 2008). By measuring the heart­rates of both Transtibial and Transfemoral amputees in
realistic walking scenarios (level ground, up/down ramp, stairs), researchers were able to quantify
via THBI differences in physiological cost energy between different when using prosthetic feet
designs required in walking. Compared to conventional energy­return prostheses, the findings
obtained from level, uphill, downhill, upstairs and downstairs extended­walking trials showed that
the new hydraulically­assistive foot offered the amputees the ability to walk with up to 8.5% less
effort.
The changes in energy management with changes is damping settings, are presented in figures 2,3
and 5. With the neutral PF­DF damping bias the range of movement at the ankle up to the point of
heel rise is approximately evenly distributed (e.g. the ankle spends nearly the same time
plantarflexing as dorsiflexing). With the brake settings in effect the PF motion duration is much
shorting demonstrating that the ankle reaches a full flat and stable condition much earlier in the gait
cycle. This is because DF motion cannot be created about the ankle until an external DF moment is
created. This finding verifies the functional requirements that when walking down an incline more
rapid movement in the PF direction is desirable to create a more stable base of support. With the
brake mode in effect the response of the foot becomes less elastic, in other words it returns less
energy thereby promoting stability and safety during stance phase.
Comparison of PF and DF damped
movement ranges
assistive bias
neutral bias
brake bias
0%
20%
40%
PF
60%
80%
100%
DF
Figure 2. Comparison between PF/DF resistance bias settings, of the damped range of movement
expressed as a percentage of full range available.
Comparison of damped motion vs
spring deflection ranges
assistive bias
neutral bias
brake bias
0%
20%
40%
Range damped
60%
80%
100%
Range Elastic
Figure 3. Comparison between PF/DF resistance bias settings, of the comparative distribution
between damped movement and spring deflection expressed as a percentage of total available
visco­elastic system movements
With the assistive mode setting enabled the results showed that the ankle spend more time
plantarflexing and that less of the total ankle motion is being transmitted through the hydraulic
system. This indicates that the ankle response during the collision is more elastic compared to other
settings. Figure 4 shows the high speed photography results indicating clear differences between
both in the positioning of the ankle unit relative to the lower carrier and the amount of heel spring
deflection that is created in each mode setting.
Figure 4. Comparison of the high speed photography results taken at the point local minimum
vertical GRF data in mid­stance, differences shown in terms of ankle and heel spring deflection for
neutral and assist modes
The fore­aft ground reaction force data (figure 6) results show that with the assist setting enabled
there is greater reactive aft force been generated at the ground interface indicating a greater
propulsive effect in comparison to the other mode settings. Interestingly also the fore shear force is
reduces indicating less breaking in the collision at the ground interface. With the brake mode
enables the aft shear is shown to be the least with proportionally the greatest fore force bias
indicating a substantially more braked collision with the ground and energy management. The
vertical ground reaction force data (figure 7) with the assistive setting enables shows a greater 2nd
peak indicating greater vertical acceleration off the force plate during late stance, suggesting a
greater propulsive effect has been created.
Summary of Key Findings
x
By adapting the hydraulics many different “virtual spring” ratings and biases can be created
enhancing the biomechanical characteristics of the foot system.
x
With the brake mode enabled (down ramp) the stability, efficiency and safety of locomotion
is increased:
o The foot reaches the flat foot condition much earlier in the GC
o The response from the foot is less elastic, and thus less reactive.
o Proportionally more braking shear at the ground interface
x
With the assist mode enabled (fast walking and up ramp) the efficiency of the locomotion
task is improved:
o More energy is directed through the spring the response becomes more elastic
o Greater aft shear indicates more propulsion in the direction of progression
o Greater 2nd peak on the vertical ground reaction force indicates more vertical
acceleration of the body mass during push­off.
Figure 5. Comparison of ankle motion data with different damping settings the magnitude of the
peaks and timing of the peaks provide some indication of changes to the mechanical response,
shorter peaks indicate that more of the total ankle movement is being directed through the heel
springs. Larger peaks indicate porportionaly more relative motion is being transmitted throught the
hydraulic system. In the legend lower settings equate to lower damping resistance settings.
Figure 6. Fore­aft ground reaction force data with the each of the 3 mode settings enabled, the
greatest aft shear is shown with the assistive mode enabled, whilst proportionally a greater fore
shear indicates greater braking shear with the brake mode enabled.
figure 7, Vertical ground reaction force date with each of the 3 mode settings enabled. The larger 2nd
peak with the assist mode active suggests greater vertical acceleration of the body mass during push
off, enhancing the push off and step to step transition dynamics
Conclusions
The conclusion from this study supports the view that future ankle­foot systems should ideally have
the capability to adjust to better optimize the gait tasks being undertaken. The consequences of
pathological kinetics at the ankle we believe contribute greatly to the energetics of locomotion and
the degree to which amputees have to adapt and compensate their gait. In more demanding
locomotion tasks the degree of compensation appears to be much greater, this is evident in the
existing THBI data and future studies already underway. We conclude that the interaction of the
foot with the environment and varied loading requirements experienced by prostheses for different
walking scenarios and conditions requires a step wise change in prosthetic design philosophy to
normalise locomotion energetic as single mode device behaviour adds greatly to the costs and
demands of prosthetic use. From our results to date we have demonstrated that it is possible to
actively adapt the rolling collision of the foot with the ground interface in a way that has a positive
effect biomechanically for different walking tasks. Future studies are underway to provide more
insight into the muscular control effects on the part of the amputee and more widely into integrated
multi­joint prosthetic control systems.
References
Portnoy S, et al. (2011) Outdoor dynamic subject­specific evaluation of internal stresses in the
residual limb: Hydraulic energy­stored prosthetic foot compared to conventional energy­stored
prosthetic feet. Gait & Posture, Volume 35, Issue 1, Pages 121­125
Khadra, H. (2008). Pilot Study for the Evaluation of a Novel Prosthetic Foot Design with Viscoelastic
Properties. MSc thesis, University of Surrey, Guildford, United Kingdom.
Moser.D et.al. (2009) Biomechanical analysis of a novel automatically self­aligning ankle­foot
prosthesis Orthopädie­Technik Quarterly, English edition III/2009
Moser.D et.al (2008) US Patent 7985265