Enhancing Functional Electrical Stimulation for
Emerging Rehabilitation Robotics in the
Framework of Hyper Project
F. Brunetti∗,∗∗, A. Garay∗∗, J.C. Moreno∗and J.L. Pons∗
∗Bioengineering Group, CSIC, Spain
∗∗Catholic University of Asunci´ on, Paraguay
Abstract. This paper presents the development of a novel functional
electrical stimulation (FES) system. New approaches in emerging reha-
bilitation robotics propose the use of residual muscular activity or limbs
movements during the rehabilitation process of neuromotor. More ambi-
tious projects propose the use of FES systems to restore or compensate
motor capabilities by controlling existing muscles or subject limbs. These
emerging approaches require more sophisticated FES devices in terms of
channels, signals controls and portability. In the framework of HYPER
project, such devices are being developed to support the main objective
of the project: the development of neurorobots and neuroprosthetics to
restore functional motor capabilities in patients who suffered cerebrovas-
cular accidents or spinal cord injury. The presented portable FES system
includes novel electrostimulator circuits and improved channel switching
capacities to enable emerging approaches in rehabilitation robotics.
Cerebrovascular accidents (CVA) is the second leading cause of death worldwide,
the third one in industrialized countries. In fact, it is considered as the main cause
of long-term disability causing a tremendous economical and societal burden.
Each year, about 795,000 people suffer a stroke in US. Nowadays, stoke death
rate is about 13.5%. Approximately the 75% of all strokes occur in people over
the age of 65 causing important motor impairments, affecting their quality of
life, and struggling the chance of independent living.
Spinal Cord Injuries (SCI) can be considered as the second leading cause
of long term motor disabilities. In numbers, approximately 11,000 new injuries
occur each year in US. In the EU, an estimated 180,000 to 230,000 people suffer
from spinal cord injury and there is an incidence of about 12,000 new cases
each year. Unlike CVA, almost the 50% of SCI population is between the ages
of 16 and 30, causing an even deeper impact in social structure and important
economical burdens, . Together with physiological and pathological tremors
, CVA and SCI are the most important causes or motor impairments.
In many cases, motor capabilities affected by CVA can be rehabilitated. The
use of robotics systems for this goal is almost a common approach in devel-
oped countries. Current approaches are focused on robotic exoskeletons that
2 F. Brunetti, A. Garay, J.C. Moreno and J.L. Pons
replicate the characteristic movements of a rehabilitation session driven by a
therapist. In this approach, potential capabilities of smart robotic systems are
not exploited, since the role of the robot is very passive, and moreover, the
human-robot cognitive interaction is scarce. In the case of SCI, rehabilitation
is quite more complicated. Many scientific works are aimed at developing solu-
tions for the well-known problem of neural tissue and motor function restoration
. Cutting edge approaches propose brain-neural computer interfaces (BNCI)
as an integrated tool to get a deep and natural human-robot interaction, both
physical and cognitive, to expand the limits of neuromotor rehabilitation after
CVA, , or to compensate neuromotor disorders like tremor, . These novel
human-robot interaction experiences include augmented reality to motivate user,
bioinspired and optimal actuators to facilitate the rehabilitation process or to
soft the physical interface and make the exoskeleton lighter and portable, among
other new features.
In this context, Hyper project was conceived. Hyper is an initiative of sev-
eral spanish research centers in field of robotics, biomechanics, neurology, brain-
computer and human-computer interfaces, advanced power management and
battery systems for portable devices. The project intends to represent a break-
through in the research of neurorobotic and motor neuroprosthetic devices in
close cooperation with the human body, both for rehabilitation and functional
compensation of motor disorders in activities of daily living. The main objectives
of the project are to restore motor function in SCI patients through functional
compensation and to promote motor control re-learning in patients suffering
from CVA by means of an integrated use of neurorobotics and neuroprosthetics.
The project will combine biological and artificial structures in order to overcome
the major limitations of current rehabilitation solutions for the particular case of
CVA and SCI. The project addresses key questions at the frontier of knowledge
in various scientific and technological disciplines. these questions are investi-
gated in six research tracks with horizontal interrelationships: the systems will
deal with variability in the human neuromuscular structures, with dynamical
adaptations according to the latent motor capabilities of the users.
The ambitious objectives of Hyper can only be achieved with the support
of adequate technologies. In this matter, the project dedicates efforts to specific
technological developments, such as improved batteries, exoskeletons, sensors,
control architectures, and novel actuators. Moreover, Hyper project collaborates
with other satellite projects like TERERE, which is aimed at the development
of electronic devices and systems for emerging rehabilitation robotics.
The work presented in this paper is focused on development of a Functional
Electrical Stimulation system in the framework of the Hyper. Previous experi-
ences in wearable exoskeletons demonstrated the difficulty of getting adequate
performance of classic actuators in terms of power, portability and weight, in
order to replace, replicate or compensate the function of human muscles, .
Hyper looks forward to contributing to the state of the art in modern actua-
tors. These actuators should enable the development of the proposed portable
exoskeletons. In this regard, the use of existing actuators is crucial. The mus-
Enhancing Functional Electrical Stimulation3
cles, that cannot be controlled anymore due to the motor impairment, will be the
robotic actuators. By means of advanced BNCI and novel electrostimulators, the
users could control their own muscles to provide the needed torque to joints to
generate the desired forces and movements of limbs. However, to achieve this is
a reliable manner, new FES systems should be able to tackle current limitation
of the technologies, such as the provoked fatigue and coarse movements.
The paper provides a short review of FES technology and existing devices,
and presents the enhanced solution designed in Hyper project.
2 Functional Electrical Stimulation Systems
Functional Electrical Stimulation (FES) consists in the excitation of muscle fibers
or nerves by means of electric impulses to provoke a movement of organs or limbs.
The electrodes used as an interface between the tissue and the electric impulses
source can innervated the muscles, be attached to a nerve or placed externally
over the skin. Thus, these systems can be invasive or non-invasive. The presented
development was designed considering a non-invasive approach.
FES is not a novel technique, but it was rather restricted to labs as a research
tool, or to clinics for rehabilitation sessions of paralyzed limbs. In 1985, the use
of FES for restoring walking in SCI patients was already proposed by Bajd et al.
, . Later, some researchers combined FES with orthoses or exoskeletons to
compensate the coarse movements provoked by existing FES technologies, .
Currently, FES is a more common and accepted technology. In the literature,
many applications can be found such as the compensation drop foot after CVA,
[?]. Veltink et al. developed smart sensors to be integrated into a implantable
FES device to for drop-foot, [?]. The trend in walking restoration after SCI
clearly points toward the use of FES systems in combination with modern ex-
oskeletons, , . However, classic FES problems remain in most of these
application, such as user fatigue, , poor energy efficiency, lack of autonomy,
scarce selectivity, of muscles, , and limited joint torques exerted by FES. By
means of more optimal FES systems in combination with external light exoskele-
tons actuators, Hyper plans to tackle all the commented problems.
2.1 Key aspects of FES
FES technique directly interacts with living tissues. A FES system must know
and respect the limitations and characteristics of the tissue. Otherwise, com-
mon problems will rapidly appear, like tissue damage, pain, fatigue, interface or
electrode corrosion and poor selectivity.
The first decision that designers of non-invasive FES devices have to make
is the type of signal that it is going to be applied to the skin. The delivered
energy to contract the muscles can be in the form of current or voltage pulses.
At first sight, both have the same effect. However, based on the neurophysiol-
ogy of muscles current pulses look more natural, since that neural signal path
4F. Brunetti, A. Garay, J.C. Moreno and J.L. Pons
is based of electric charges transmission. Moreover, in transcutaneous or non-
invasive FES systems, the interface impedance can easily change and provoke
different contraction levels that directly depend on the electric charges applied
to muscle fibers. This is avoided if current pulses are applied. In terms of elec-
tronic design, a current source capable of handling the skin-electrode impedance
is more complex to obtain, even more when considering a portable design for
wearable rehabilitation robotic scenarios.
Another very important issue is the waveform of the applied current pulses.
Figure 1 depicts several waveforms that have been studied, . Waveforms can
be monophasic (a,b) or biphasic (c, d, e, f). Some FES systems apply balanced
charge biphasic stimulation patterns, which means that the applied energy to
skin, in terms of charges, is the same during the positive and negative cycles, as
occurs in waveform (c, d, e, f). It has been demonstrated that balanced waveform
reduces the muscle fatigue. Regarding the shape of the waveforms, most of the
depicted ones are based on rectangular pulses. Studies demonstrated better cell
and fiber, , recruitment results comparing to triangular and other functions.
The electrostimulated plant (electrodes-skin-muscle) can be modeled as an
electric resistance in parallel with a capacitor, and this arrangement in series
again with another resistance, . If rectangular voltage pulses are applied, RC
charge-discharge curves appears in current waveforms, affecting the recruitment
of muscle fibers and cells. This justifies the use of a current source instead of a
Frequency, amplitude and duty cycle are other aspects that have to be con-
sidered when designing a FES device. Moderns electronic devices allows the
fulfillment of most requirements in this regard. Amplitude of current signals
can vary between 10 mA and approximately 150 mA depending on the mus-
cle groups stimulated, the electrode-skin interface and the pain threshold of the
user. Frequency range between 5 and 50 Hz. Muscle fatigue is also related with
the frequency. It increases with frequency. Regarding the duty cycle, rectangular
pulse width is usually between 100 µs and 3 ms, .
2.2 Open Challenges
Modern FES points towards the use of distributed electrodes to stimulate mus-
cles. This approach is twofold. First, it is desired to reduce the amount of the
current used to contract the muscles at each electrode-skin contact point. This
would reduce pain and fatigue. Second, with more contact points and electrodes,
muscle selectivity can be improved, .
The distributed concept also demands new stimulation algorithms and pat-
terns to reach the desired goals. In this matter, flexible waveforms generators and
multichannel systems are required. Some previous works have proposed tech-
niques for optimization of spatial selectivity of multi-pad electrodes,  and
their potential application in enabling functional movements, .
These former requirements are directly related with the electronic design of
the FES device. Voltages at the electrode-skin interface can easily reach above
100V considering a 100 mA current pulse and an impedance of 1 kΩ. Distributed
Enhancing Functional Electrical Stimulation5
0 0.002 0.0040.006 0.0080.01
Common FES patterns
0 0.0020.004 0.0060.008 0.01
0 0.002 0.0040.0060.0080.01
0 0.002 0.004 0.006 0.008 0.01
0 0.0020.0040.0060.008 0.01
0 0.0020.004 0.0060.0080.01
Fig.1. Common current waveform used in FES systems. X-axis correponds to
time represented in seconds.
electrodes mean dozens of electrodes that a have to be connected to the stimu-
lator and moreover stable current stimulator cannot be obtained so easily using
compact electronic components unless using a complete microelectronics system
design, in which the power management and dissipation could be one of the
Emerging rehabilitation robotic solutions, like the one proposed by Hyper
project, need flexible, portable, controllable, robust and multichannel FES de-
vices. This can only be achieved by means of new advances in portable current
sources with high voltage electronic switching.
6 F. Brunetti, A. Garay, J.C. Moreno and J.L. Pons
Fig.2. Basic biphasic transconductance amplifiers or voltage controlled current
The most basic electrostimulator circuit is the one shown in figure 2. It can
provide a monophasic or biphasic output waveform, similar to those shown in
figure 1 (a), (b), and (c).
The amplitude of the current pulses is proportional to the amplitude of the
voltage V1 and V2 applied to base pin of the transistor near the voltage source.
The symmetric arrangement of two transistors, the one connecter to the voltage
source and the other connected to collector the former one, work as a transcon-
ductance amplifier. The other two transistors act as single switches. The problem
with this simple circuit is that it works in an open loop fashion and the transcon-
ductance gain, gm, depends on each transistor and cannot be controlled.
In a previous section, the importance of balanced charge biphasic waveform
was stated. To achieve this goal, Keller proposed a two staget circuit with two
different working phases, . During the first phase, the phase of active stimu-
lation of the system, an operational amplifier controls the current across the load
connected to a high voltage source. During the second phase, a passive discharge
circuit, which appears in the upper part of figure 3, removes the electric charges
from the electrodes-skin-muscle interfaces, previously inserted during the active
phase. This is a very interesting solution but it struggles the chance of modifying
actively the waveforms of the discharging periods.
Enhancing Functional Electrical Stimulation7
Fig.3. The biphasic current source circuit with balanced charge proposed by
Thierry Keller, .
2.4 Existing solutions
There are many FES devices available on the market. However, as a result of
an extensive literature survey, three models, two of then portable, are the most
adequate for rehabilitation robotics and scientific purposes.
The first one is the Compex Motion, developed by T. Keller et al. in 2002.
The Compex Motion was developed to control a hand neuroprosthesis, . It
is a current controlled stimulator, with a current range between 0 and 120 mA,
pulse width between 0-16 ms, and a pulse frequency up to to 100 Hz. The device
provides four channels for stimulation and some other digital and ADC inputs.
The battery is rechargeable. The inventors claim that monophasic and biphasic
waveforms can be obtained, .
The second reviewed model is the UNAFET8, . This device provides a
USB connection to connect to the PC. The Compex Motion provided a RS-
232 interface. The UNAFET is similar to the Compex since it provides a current
controlled output. The voltage output can be up to 150 depending of the electric
load. The pulse amplitude goes from 0 to 50 mA and the pulse width can be
set between 0 and 1 ms. The frequency rage is between 5-80 Hz. The inventors
refer to the output as monophasic charge compensated. This means a waveform
output similar to that shown in figure 1(e) and it is similar to the provided by the
Compex Motion device. They probably called it this way because the discharging
period is achieved passively, and not by the current source. The main advantage
of this device comparing to the Compex Motion is the number of channels that
in the case of the UNAFET8 is eight.
The third and last reviewed FES device is the RehaStim. It was developed by
Hasomed GmbH, . It can stimulate up to 8 independent channels. The current
amplitude ranges from 0 to 126 mA. The waveform is biphasic. The frequency
8 F. Brunetti, A. Garay, J.C. Moreno and J.L. Pons
Fig.4. The TEREFES block diagram.
range is between 0 and 180 Hz. The communication interface is based on USB
technology. This FES device is the most flexible one, but it is not portable.
3 The TEREFES design
The design proposed in the framework of Hyper and TERERE projects look
forward to providing a high number of electrostimulation channels driven by
controlled and stable current sources. The design has to be portable and flex-
ible. This means that the generated waveforms could be modified to explore
novel stimulation algorithms and patterns to support this emerging rehabilita-
tion robotic field.
The proposed architecture Is depicted in figure 4 and its called TEREFES.
Four AA batteries power the device. It includes USB communication interface
by means of a FTDI232BL IC. This allows the configuration by an external soft-
ware application. The stimulation circuit is controlled by an Atmel Atmega128
microcontroller. All the digital circuit is isolated and digital to analog convertors
are used to control, by means of a voltage signal, the amplitude of current pulses
used for stimulation.
The amplitude of the current pulses is between 0 and 120 mA (256 steps),
with a maximum voltage of 250 V. The frequency can be configured with values
between 0 and 100 Hz. It provides up to 32 channels, in two independent groups
of 16 channels each, and driven by two different electrostimulators. The pulse
width can vary from 10 to 5,000 µs. It also includes some general purpose TTL
digital I/O tfor synchronization with other devices and if required for external
sensors signal acquisition.
There are two main components in the novel device: the electrostimulators
and the switching circuits.
Enhancing Functional Electrical Stimulation9
Fig.5. The basic circuit of the TEREFES electrostimulator.
Hyper proposes a more stable current source acting as the core of the novel
FES device. The source current is based on a close loop voltage to current,
or transconductance, amplifier. The basic circuit diagram is depicted in figure
5. A high voltage operational amplifier, the APEX PA78, enable this circuit.
Within the proposed circuit, by choosing adequate values of R1 , R2 and Rs, the
transconductance gain, gm, can be set.
The microcontroller provides a low voltage signal that is transformed to a
current signal that crosses along the load. By proper switching of the loads, fully
controlled monophasic and biphasic signals can be obtained. Since it is current
source, when the electric load is high, meaning an open circuit, the voltage of the
electrodes reaches its maximum value (250 V). Special techniques were developed
to minimize this voltage when switching the loads to obtain biphasic waveforms.
The technical specification of the high voltage OPAMP allow fast rising times.
Comparing to the two different current sources previously reviewed, the basic
circuit based on transistors and the Keller’s approach, the Hyper proposal has
two main advantages. First, the gmvalue remains constant within the range of
operation. The second advantage is that the load is connected at a 0 V point
when the controlling signal is equal to cero. This does not happen in the circuit
depicted in figure 3. In this case, the load is always connected to a high voltage
The second key component of the TEREFES design is the switching circuit.
The aim of this circuit is to provide in a small form factor the capabilities of
applying a unique excitation signal source to multiple electrodes. The solution
comes from the field of microelectronics for piezoelectric arrays. The TEREFES
uses the MAX14802 that is a 16-channel high voltage analog switch IC, controlled
by a serial bus. The microcontroller generates the adequate signals to control the
10 F. Brunetti, A. Garay, J.C. Moreno and J.L. Pons
Fig.6. Basic switching circuit of the TEREFES electrostimulator. The switches
are part of the MAX14802, a 16 high voltage anlog switches IC.
used scheme, depicted in figure 6, enabling the implementation of monophasic
and biphasic waveforms. The TEREFES uses 4 ICs to get the full functionality of
two stimulator with 16 channels each. The configuration uses a common cathode
scheme for each 16-channel group. The setup is adequate to control two different
muscle groups in a distributed manner simultaneously.
4 Implementation and results
Implementation of TEREFES design resulted in a 190 x 138 x 45.5 cm device.
The weight is approximately 300 grams. The autonomy could be measured under
different conditions, but early results indicate an autonomy of approximately 4
hours under normal conditions. The complete design is divided in three different
interconnected printed circuit boards. The first one includes the power circuit.
The second one includes the digital controller, isolators, and switching circuit.
The high voltage operational amplifiers are located in different printed board
Figure 7 depicts a basic high voltage waveform obtained with the TEREFES
electrostimulators and the switching circuits with a pure resistive load.
Enhancing Functional Electrical Stimulation11
Fig.7. Basic waveforms obtained with a pure resistive load and the TEREFES
5 Future work
In the framework of HYPER project the system will be validated with normal
subjects and with CVA and SCI patients. Ad-hoc designs will be implemented
to get optimal configuration in terms of number of channels and stimulators.
The TEREFES is not only a flexible FES device, but also a modular architec-
ture. Future works will be focused on the development of alternative switching
modules in order to get different configurations, resulting from the trade-off
between the number of channels and muscle groups. For instance, for other re-
habilitation robotics scenarios, a more adequate configuration of the TERFES
could be one with two stimulators but for eight different muscle groups, each one
with up to four electrodes. This would require eight different cathodes, unlike
the current configuration, which has only two.
This paper presents a novel FES device system. The system was conceived in
the framework of Hyper project, which proposes ambitious rehabilitation neuro-
robotic and neuroprosthetic systems for SCI and CVA patients.
The developed electrostimulator is based on a novel current source that rep-
resent a step forward in current state of the art in FES devices. It represents a
robust designed based on a linear closed loop transconductance amplifier.
The FES system, called TEREFES, also provides enhanced switching ca-
pacities to support multiple electrodes. In the described design with up to 32
channels are obtained in a portable device. The electrostimulator can drive in
each channel a 120 mA biphasic waveform.
12 F. Brunetti, A. Garay, J.C. Moreno and J.L. Pons
The authors would like to thank to HYPER consortium. HYPER project is
founded by the Spanish Ministry of Science and Innovation under the program
CONSOLIDER-INGENIO 2010. TERERE project is founded by the Spanish
Agency for International Cooperation for the Development.
1. Lim, P., Tow, A.M.: Recovery and regeneration after spinal cord injury: A review
and summary of recent literature. Annals Academy of Medicine 36(1) (2007) 49–57
2. Rocon, E., Belda-Lois, J.M., Snchez-Lacuesta, J., Pons, J.L.: Pathological tremor
management: modelling, compensatory technology and evaluation.
Technology and Disability 16 (2004) 3–18 1
3. Dietz, V.: Ready for human spinal cord repair? Brain 131 (2008) 2240–2242 1
4. Moreno, J.C., Lagoda, C., Giovanni, A.D., Torricelli, D., Pons, J.L.: Brain-neural
computer interaction for evaluation and testing of physical therapies in stroke
rehabilitation of gait disorders. In: Proccedings of Beyond Brain Machine Interfaces
Workshop: From Senses to Cognition. (2010) 1
5. Gallego, J., Rocon, E., Ibez, J., Koutsou, A.D., Pons, J.L.: A soft wearable robot for
tremor assessment and suppression. In: Proceedings of the 2011 IEEE International
Conference on Robotics and Automation. (2011) 1
6. Cullell, A., Moreno, J.C., Rocon, E., Forner-Cordero, A., Pons, J.L.: Biologically
based design of an actuator system for a knee-ankle-foot orthosis. Mechanism and
Machine Theory 44 (2009) 860–872 1
7. Bajd, T., Andrews, B.J., Kralj, A., Katakis, J.: Restoration of walking in pa-
tients with incomplete spinal cord injuries by use of surface electrical stimulation
- preliminary results. Prosthetics and Orthotics International 9 (1985) 109–111 2
8. Bajd, T., Kralj, A., Turk, R., Benko, H., Sega, J.: Use of functional electrical
stimulation in the rehabilitation of patients with incomplete spinal cord injuries.
Journal of Biomedical Engineering 11(2) (1989) 96–102 2
9. Andrews, B., Baxendale, R., Barnett, R., Phillips, G., Yamazaki, T., Paul, J.,
Freeman, P.: Hybrid fes orthosis incorporating closed loop control and sensory
feedback. In: The 27th Annual Scientific Meeting of the Biological Engineering
Society. (1987) 2
10. Quintero, H., Farris, R., Durfee, W., Goldfarb, M.: Feasibility of a hybrid-fes system
for gait restoration in paraplegics. In: 2010 Annual International Conference of the
IEEE Engineering in Medicine and Biology Society (EMBC),. (2010) 2
11. Pons, J.L.: Rehabilitation exoskeletal robotics. IEEE Engineering in Medicine and
Biology Magazine May-June (2010) 57–63 2
12. Kobetic, R., Curtis, S., Schnellenberger, J., Audu, M., Bulea, T., Gaudio, R.,
Pinault, G., Tashman, S., Triolo, R.: Development of hybrid orthosis for standing,
walking, and stair climbing after spinal cord injury.
Research & Development 46(3) (2009) 447–462 2
13. Koutsou, A.D., Moreno, J.C., Pons, J.L., Gallego, J.A., Rocon, E.: Muscle selec-
tivity algorithm for superficial matrix electrodes. In: 9th International Symposium
Computer Methods in Biomechanics and Biomedical Engineering. (2010) 2, 2.2
Journal of Rehabilitation
Enhancing Functional Electrical Stimulation13 Download full-text
14. Durand, D.M.: 17. Electric Stimulation dfo excitable tissue. In: The Biomedical
Engineering Handbook. CRC Press (2000) 1–23 2.1
15. McNulty, M., Fogarty, P.: Design of a highly efficient circuit for electrical muscle
stimulation. In: IEEE Biomedical Circuits and Systems Conference. (2006) 2.1
16. Thrasher, T., Flett, H., Popovic, M.: Gait training regimen for incomplete spinal
cord injury using functional electrical stimulation. Spinal Cord 44 (2006) 357–361
17. Malesevic, N., Popovic, L., Bijelic, G., Kvascev, G.: Classification of muscle twitch
response using ann: Application in multi-pad electrode optimization. In: The 10th
Symposium on Neural Network Applications in Electrical Engineering. (2010) 2.2
18. Lawrence, M., Gross, G., Lang, M., Kuhn, A., Keller, T., Morari, M.: Assessment of
finger forces and wrist torques for functional grasp using new multichannel textile
neuroprostheses. Artificial Organs 32(8) (2008) 634–638 2.2
19. Keller, T.: Surface functional electrical stimulation neuroprostheses for grasping.
PhD thesis, ETHZ (2001) 2.3, 3
20. Popovic, M., Popovic, D., Keller, T.: Neuroprosthesis for grasping. Neurological
Research 24(5) (2002) 443–452 2.4
21. Keller, T., Popovic, M., Pappas, I., Mlle, P.: Transcutaneous functional electrical
stimulator ”compex motion”. Artificial Organs 26(3) (2002) 219–223 2.4
22. Broderick, B., Breen, P.P., Laighin, G.: Electronic stimulators for surface neural
prosthesis. Journal of Automatic Control 18(2) (2008) 25–33 2.4
23. Schauer, T., Negaard, N., Behling, C.: Sciencemode, rehastim stimulation device.
Technical report, Technische Universitt Berlin, Control System Group (2009) 2.4