Hugh Herr, PhD |
The science and technology research accomplishments by Hugh Herr,
PhD, have already had a significant effect on physically challenged
people. The Variable-Damper Knee Prosthesis has
recently been commercialized by Össur Inc., and is now benefiting
transfemoral amputees throughout the world. In addition, the Active
Ankle-Foot Orthosis is now in the process of being commercialized,
and has the potential for improving the quality of life of millions
of stroke patients in the U.S. alone. Professor Herr has given numerous
lectures at international conferences and colloquia, including the
IVth World Congress of Biomechanics, the International Conference on
Advanced Prosthetics, the National Assembly of Physical Medicine and
Rehabilitation, the Highlands Forum XXII (Life Sciences, Complexity,
and National Security), and the TEDMED International Conference. He
is Associate Editor of the Journal of NeuroEngineering and Rehabilitation,
and has served as a reviewer for the Journal of Experimental Biology,
the International Journal of Robotics Research, IEEE Transactions on
Biomedical Engineering, and the Proceedings of the Royal Society: Biological
Sciences.
|
|
It´s an honor to be here.
In my research program, we think about the biomechanics and control
of the whole organism—the whole body—down to the individual
muscle cell, and we apply these sciences to the development of various
technologies: prosthetics, orthotics, and also exoskeletons for human
augmentation.
|
|
|
|
Today, I´m going to speak primarily about leg prosthetics, but
a lot of what I have to say is also appropriate for upper-extremity amputees
as well.
I´ll begin with a project that has been underway in my laboratory
for many years, and it has been completed. It´s now a product.
And then I´m going to spend the rest of the time making fun of
that device and telling you what´s wrong with it and what we need
to do as a community.
The device is a transfemoral external knee prosthesis. We call it the
Rheo Knee™ System (Össur, Reykjavik, IS), because it uses
magnetorheological fluid.
|
|
|
|
How does it work? The device contains carrier oil and small iron particles
suspended in that carrier fluid. We modulate the magnetic field inside
the knee, and by doing so, we can vary knee resistance, or damping, quickly
and quietly. So, this is a variable damping technology.
We recently compared the Rheo™ to two systems that have been
on the market for quite some time: the Mauch S&S® (Mauch, Dayton,
OH), a passive hydraulic system, and the C-Leg® (Otto Bock, Minneapolis,
MN), which we all are familiar with. The C-Leg® is also a hydraulic
unit, like the Mauch, but it´s controlled by a microprocessor.
|
|
|
|
In the study, we found that the amount of food energy the above-knee
amputee requires to move from point A to point B is affected by these
distinct knee designs. We got a small but significant effect with the
Rheo™, a reduction in metabolic cost or an increase in walking
economy.
|
|
|
|
We´ve looked into the biomechanical mechanisms that might explain
this difference, and what we´ve found is that the Rheo™,
because of its different strategy for developing knee resistance or damping,
is able to reduce the muscular effort at the hip on the affected side.
In terms of control, the Rheo™ knee is fairly adaptive. It often
does not require a human to program the knee to the patient. The knee
adapts its damping parameters to the patient, allowing the patient to
walk at different speeds and across different terrains. The Rheo™ essentially
does experiments and optimizes itself to the patient.
|
|
|
|
Patients with a dumb knee, with no computational intelligence whatsoever,
no sensors or whatnot, will have a highly pathological gait for ascending
or descending stairs. This is true even for patients who have adapted
quite well, whose unaffected, biological limb is quite strong.
With the Rheo™, patients are able to walk up steps foot-over-foot.
The knee recognizes that patients are going up steps and outputs the
appropriate algorithm. Patients are also, of course, able to go down
steps.
Recently, Össur, the second-largest prosthetic manufacturer in
the world—they´re in Iceland—launched the Rheo™ as
a product. It´s now available to amputees throughout Europe and
the United States.
Moving on, now I´m going to make fun of what I just told you
about. What´s the opportunity for making advances in this area?
What are the issues?
Commercially available above-knee prostheses are variable damping mechanisms.
What do I mean by this? I call them fancy car brakes. All they´re
able to do is dissipate mechanical energy.
The human knee is capable of dissipating mechanical energy, but additionally
it´s capable of actually supplying a motive force or torque, and
also acting like a spring and varying its stiffness. We need to do better
in the prosthetic knee department.
In the foot/ankle department, current commercially available systems
are completely passive, typically spring, devices. In a flex-foot design,
you don´t see an actuator or sensors. Therefore, the system has
no ability to adapt to the amputee.
|
|
|
|
Biologists tell us there´s a lot going on in the healthy human
ankle. During the early stance phase in level-ground walking, even at
a constant walking speed, the stiffness of the ankle is constantly being
updated by the central nervous system. In late stance, the ankle supplies
a tremendous amount of positive power, a motive moment, which is believed
to be very important to human ambulation.
Given the passive nature of today´s commercially available prostheses,
how does that affect the patient? It causes a pathological gait, a limp,
which typically causes excessive impact forces to the musculoskeletal
system, which can trigger difficulties later in life, i.e., back problems
and whatnot.
Normally, amputees also require a greater amount of food energy to
go from point A to point B, substantially greater. The Rheo™ knee
improved that somewhat, but we have a great deal to go from here.
To really push this area of medicine, we need to merge body with machine,
to create an intimacy between the human body and the prosthetic device.
In the interest of time, I´d like to describe two key areas,
although other critical issues exist.
First, we need better motor systems, better actuators that are muscle-like.
Second, we need distributed sensing and intelligence.
|
|
|
|
Beginning with muscle-like actuation, why are muscles so fabulous?
Why do we desire to have muscle-like actuators? Muscle tissue has excellent
functional characteristics. It´s very mechanically powerful given
its size. You can typically get 50 watts per kilogram of muscle tissue
for continuous operation.
And muscle is functionally adaptive. We all know this. If we´re
couch potatoes and we do not exercise, then our muscles become weak.
But if we work out, they scale to the task.
uscles are very scalable. They´re in small critters all the way
up to blue whales. Muscles are quiet, too. If we were all humanoid robots
in this room, powered by a synthetic motor system, gasoline powered engine,
electric motor, whatever, we´d raise a tremendous racket and barely
would be able to communicate.
|
|
|
|
Noise is a really important concern and it´s a difficult problem
to solve. Could we use muscle tissue in our robots and prostheses? Sounds
crazy, but in my group we´re actually thinking about this idea
of hybrid devices, where part of the device is living tissue and the
remaining component is synthetic.
e´ve built this swimming robot. It´s swimming through its
own media, which comprises antibiotics, antimycotics, and also glucose
to feed the muscle tissues.
|
|
|
|
So, we might want to think that someday the prosthetic hand, for example,
would be this hybrid device where we´d use synthetic components
only where synthetic components are better than biological components.
That´s a hard problem, obviously, and it´s going to take
a few years to solve. In the short term, what do we do? Now, I want to
talk about actuator systems that are muscle-like to some degree.
everal years ago, in the field of robotics, my friend and colleague
Gill Pratt developed what´s called the series-elastic actuator. It´s
muscle-like in an abstract sense because what you have is an electric motor
in series or next to a compliant spring, kind of like a muscle belly in
series with a tendon.
|
|
|
|
To control the device, we sense the amount of energy that´s stored
in the series spring, similar to an artificial golgi tendon organ, and
the control system basically controls how much energy is in the series
spring or the spring deflection. By doing that, we can accurately control
the forces that the system exerts on the world. It´s very shock-tolerant,
and very force-controllable.
This was originally developed for legged robots. We have a dinosaur
robot we call Trudy that is autonomous, carries its own power supply,
and walks in three-dimensional space. Trudy uses these series-elastic
actuators.
Recently, we´ve also used the actuators for rehab in my group.
We had a gentleman who had suffered a stroke. He had this classic drop-foot
condition where the muscles of the anterior compartment of the leg were
weak, so he´d hit the ground on his left side with his forefoot
instead of his heel.
We developed a robot that wraps around his leg that pushes on him and
restores his gait. With the device, there´s a better symmetry between
affected and unaffected sides and he´s able to walk at a faster
speed.
That´s one possibility. But this system relies on electric motors.
Electric motors are not silent. They´re better than gasoline-powered
engines, for sure, but you can still hear them.
Electric motors also require a power supply. We´re often constrained
to use battery technology, which has a rather poor energy density.
What about artificial muscle? This doesn´t help us in the efficiency
or the transduction efficiency arena, but it may help us in terms of
the fact that artificial muscles are linear and they´re also very
quiet.
|
|
|
|
There is a series of muscles, electroactive polymers, that has been
developed by SRI International in California, by Roy Kornbluh and his
colleagues. In my view, their artificial muscle´s are extremely
impressive.
They´ve done sort of a finger-type embodiment. It´s activated
by applying high voltages. They´ve already used the muscle in biomimetic-type
robots. They also have a giant fly, where the muscles are distorting
the thorax of the machine, which flaps the wings. The fly can actually
get off the ground, but they have no idea how to control it..
What are the remaining issues? One is a scaling issue. At SRI, they´re
very good at building muscles the size of your middle finger, but to
build a gastrocnemius is more difficult.
Also, there´s a cycle life issue. If you keep the strains very
modest, just a few percent, you can get a million cycles. However, at
physiologic strains of 20 percent, the muscle breaks down quickly.
The muscle requires high voltage, as I mentioned, but if you keep the
currents very low, a human can safely interact with the muscle. This
is exciting, and I believe researchers in this area will solve these
remaining problems.
So, we have muscle-like actuators, and that´s indeed important.
But how we use the muscle actuators, the muscle/skeletal architecture,
is also critical.
It would indeed be a mistake to simply put one motor per degree of
freedom. In our body, as we all know, some of our muscles span a single
joint and some muscles span two joints and other muscles, polyarticular
muscles, span more than two joints. Biologists tell us that this is important
for having lightweight limbs, especially distally. Polyarticular actuation
is important to have muscles that are proximal that do work and exert
control distally.
|
|
|
|
I´m going to quickly take you through a walking step and give
you a sense of how this works and why it´s important. A typical
walking step has seven stages, from heel-strike to toe-off. At stage
three, the hip extends, which, since the foot is on the ground, straightens
the knee.
As the knee straightens, since we have the gastrocnemius that spans
both the ankle and the knee, and that´s linked to this massive
tendon, the Achilles tendon, that action of actively extending the hip
using hip extensors actually pumps energy into the Achilles spring. Then
that energy can, in turn, power the ankle.
This is very intriguing because we can think of an above-knee prosthesis
where we actually harness the muscles of the amputee and we use those
energies, we transfer those energies past the knee to power the ankle.
That´s very compelling because in principle one could do this with
very small motors and variable damper and passive spring systems. This
approach would lead to a low-mass, fairly quiet system. Again, it´s
not only the muscle-like actuators, but how we use them that is critical.
I´m going to finish with distributed sensing intelligence. Again,
I´m going to make fun of my own design. The Rheo™ is adaptive
and it adapts because it knows something about walking—biomechanical
knowledge—and it knows something about how prosthetists can adjust
alignment and knee resistance to get an amputee to walk better.
|
|
|
|
But the knee doesn´t have a direct measure of what the person
wants, the user intent. With the Rheo™ or the C-Leg®, or all
these systems, the amputee has no way to tell their knee that there are
stairs up ahead, or there´s a pothole.
|
|
|
|
We, too, are beginning to work with the Alfred Mann Foundation, and—as
we all know—they´ve developed this wonderful technology called
the BION®. We just heard a talk about functional electrical stimulation
( FES), where the BION® can be used to control skeletal muscle.
We can also think about the BION® as a sensor. We can implant the
BION® into muscle and measure the extent to which the spinal cord
has depolarized a muscle cell.
|
|
|
| We recently conducted an experimental session. I got wired
up, and we measured my electromyographic signals from my residual limb.
Another participant, Sam, wore a Vicon motion-capture system where we measured
the state of Sam´s leg as he moved his foot/ankle system. We´re
taking that data and trying to develop models to link the electromyographic
signals to my desired movements or biomechanics. |
|
|
|
Our plan is, about a year from now, we´ll inject BIONs® into
my residual limb, and then we´ll apply these algorithms. When I
think about moving my ankle, plantar flexing and dorsiflexing, I´ll
look down at an active ankle we´ve developed already in my laboratory
that will respond to my movement desires.
This will be very important to the amputee, who will have an active
alignment control for going up and down hills and stairs. It will also
dramatically increase the dynamic cosmesis of amputee gait.
|
|
|
|
Just a note here: Another very exciting technology that is more preliminary
and has not been fleshed out yet is the sieve electrode. One problem
that we´re going to face is this issue of an afferent sensory signal.
With the BION®, I´ll have my eyes, my visual system, to look
down and to tell me what the position of my ankle is, roughly. And we´ll
perhaps embed tactile vibration into the socket to give me an additional
afferent signal.
What would be fun is to think about the sieve electrode where we transsect
a peripheral nerve and we get it to grow through the electrode. With
this, you have bidirectional controllability, where you can actually
close the loop.
|
|
|
|
Imagine a future with this type of technology, that an amputee would
not only be able to walk across a sandy beach, but could actually feel
the sand against his prosthetic foot.
I´d like to thank my various sponsors. We are beginning to work with
the Department of Veterans Affairs (VA). In the future, the Alfred Mann
Foundation will supply us with BIONs® and engineering support. The
Defense Advanced Research Projects Agency (DARPA) is also a contributor
to this work and other projects in my laboratory. And, as I mentioned earlier, Össur,
a for-profit manufacturer of prosthetic components, helped us in the artificial
knee development.
|
|
|
|
To summarize, advances in muscle-like actuators, neuroprostheses, and
biomimetic control strategies are necessary to increase the merging of
body and machine to create an intimacy between the human body and prostheses.
It´s our thesis that such an intimacy will create a paradigm shift
in this area of medicine.
|
|
To summarize, advances in muscle-like actuators, neuroprostheses, and
biomimetic control strategies are necessary to increase the merging of
body and machine to create an intimacy between the human body and prostheses.
It´s our thesis that such an intimacy will create a paradigm shift
in this area of medicine.
|
|
|
