A Novel Closed-loop Control Approach for a

Flywheel and Electrical Clutch Assist Mechanism in

FES Cycling

S. C. Abdulla and M. O. Tokhi Department of Automatic Control and Systems Engineering

University of Sheffield

Sheffield, UK

cop11sca@sheffield.ac.uk

Abstract— A closed-loop control approach is proposed to control

a flywheel and electrical clutch mechanism to assist paralyzed

legs during functional electrical stimulation (FES) cycling by

stimulating the quadriceps muscle group. Fuzzy logic control is

used to control pedaling movement of legs by controlling the

pulse width of the stimulus. The proposed control approach with

the new assist mechanism significantly reduces the cadence error

and produces smoother pedaling movement as compared with

cycling without assistance. The results show that the new

mechanism has reduced the overall stimulation intensity on leg

muscles by approximately 43% and hence delays the occurrence

of muscle fatigue. The new mechanism promotes smooth,

coordinated and prolonged FES cycling for disabled individuals.

Keywords— Flywheel and clutch, assist mechanism, FES

cycling, closed-loop control, fuzzy logic

I. INTRODUCTION

Functional electrical stimulation (FES) has widely been

used for training paralyzed limbs. Several researches have

shown the importance of FES in improving the health of the

disabled by reducing muscle atrophy, improving blood

circulation in the limb, preventing bone loss and pressure sores,

as well as improving the self-image of the individual [1-5].

FES can be described as a train of pulses applied to the motor

units of a muscle to cause muscle contraction and generate

movement in paralyzed limbs. Stimulating the muscle with

FES can be done by using a stimulator with either implanted or

surface electrodes. Using surface electrodes is more common

because they are cheaper and safer as compared with implanted

electrodes [6]. One of the problems associated with FES is the

rapid appearance of muscle fatigue that terminates the training

session.

One of the common forms of training by FES that is widely

used in rehabilitation clinics is FES cycling which is

considered as an attractive and enjoyable choice for the

disabled due to its simplicity as compared with standing and

walking. One of the handicaps that are encountered during FES

cycling is the two dead points, i.e. a point of transition between

knee extension and flexion moment [6], that cause rapid

change in the speed and consequently unsmooth, i.e. jerky,

pedaling movement.

To provide a comfortable, smooth and prolonged exercise

for paraplegics researchers have focused on improving the

mechanical design of the cycling ergometer [6-9] and the

closed-loop control approach [10-13] as well as stimulating

different combinations of muscle groups [7][10][14]. As an

assist mechanism, some researchers used an auxiliary motor

[14-15] to help passing the cycling dead spots as well as assist

the leg pedal when muscle fatigue occurs. Others used a crank

arm to provide hybrid training as well as help the leg passing

the dead spots [6]. Others utilized an elastic spring to provide

assistance [16]. In a previous work [17], a novel assist

mechanism, represented by a flywheel and electrical clutch, has

been introduced. The flywheel, as energy storage device, is

engaged with the bicycle shaft by the clutch to absorb the

excessive energy in the system and store it as kinetic energy.

Also it re-engages with the shaft to discharge its energy into the

system and speed up the movement.

In this work a novel closed-loop control approach is

proposed to control the engagement and disengagement of the

flywheel by the clutch in FES cycling. A comparison between

FES cycling, by stimulating the quadriceps muscle group,

with/without the new assist mechanism is introduced and

assessed.

II. SYSTEM DESCRIPTION

A. Humanoid-bicycle model

A humanoid-bicycle model, with a flywheel and an

electrical clutch mechanism, was developed using Visual

Nastran 4D dynamic simulation software. The humanoid

model was built using the standard anthropometric dimensions

introduced by [18]. The dimensions of the bicycle used were

taken from a real cycling ergometer. Standard ball bearings

friction was added to crank shaft holder of the bicycle for more

realistic results. The electrical clutch mechanism was modeled

as an on/off constraint between the flywheel and the crank

shaft of the bicycle using the same software. Detailed Sponsored by The Ministry of Higher Education and Scientific Research/

KRG-IRAQ.

information about the humanoid-bicycle model can be obtained

from previous work of the authors [17].

B. Muscle model

The human musculoskeletal muscles, that are responsible

for producing voluntary movement, have been widely

described in the literature [19-20]. In this work, it is preferred

to use the model featured in [21], which is simple to implement

and accurate enough as it has been derived from data obtained

experimentally from paraplegics and healthy subjects. The

model is represented by a single-pole transfer function that

mimics the behaviour of the quadriceps muscle group

stimulated by an electrical stimulus. This muscle model is

implemented using simulink/matlab software with fixed

stimulation frequency of 33Hz, while the pulse width is kept

variable. The resultant muscle torque is fed to the knee joints of

the humanoid-bicycle model to produce leg movement. Further

information about this muscle can be found in detail in [21].

III. CONTROL STRATEGY

The amount of torque, generated by applying FES signal on

leg muscles, can be controlled by a closed-loop control

approach to maintain smooth and coordinated pedaling

movement. The controllers can be used to alter, according to a

feedback signal, the pulse width of FES signal, control the

amount of muscle contraction and produce the required

movement. In this work, it is aimed to perform a cycling

cadence of 35 rpm, the lowest cycling speed used clinically

[10]. The actual cycling speed is measured using a speed

sensor, in the humanoid-bicycle model, located at the bicycle

crank, i.e. shaft, to measure the angular velocity of the crank.

This feedback signal is compared with a reference of 35 rpm

and the resultant error signal is fed to the controllers to

accordingly alter the pulse width and control the movement of

the legs to follow the reference.

A. FES cycling without assistance (Scenario I)

The aim is for paraplegics to perform FES cycling, by

stimulating the quadriceps muscle group of each leg, without

assistance. The difficulty in this scenario arises from the fact

that stimulating the quadriceps of each leg leads to knee

extension only and no flexion movement can be generated.

Although, at certain positions, knee extension can speed up the

pedaling movement, at other positions it can also produce

opposite action and retard the movement, as in Fig. 1. To

govern a required cadence, the muscle should be stimulated

twice per cycle; first to speed up the movement and second to

retard it if necessary [16]. The cycling movement of each leg

can be divided into three phases; push, resist and rest. The push

phase is the period at which the knee extension can speed up

the cycling, while the resist phase is the period at which the

knee extension produces resistance to the movement. The rest

phase is the phase at which the leg is left without stimulus to

get rest before the next stimulus is due.

For simplicity, the stimulation phases are defined according to

the crank angle of the bicycle measured by a position sensor.

Considering the 0° of the crank as the horizontal point at which

the right leg is situated close to the hip, the phases are defined

as; right push phase between 20° and 80°, the right resist phase

between 150° and 200°, the left push phase between 200° and

260°, the left resist phase between 330° and 20°. Since the

amount of pushing required to speed up the cycling may differ

from that of resist, four fuzzy logic controllers are used as

shown in Fig. 2.

The cadence error and its derivative are fed to the

controllers after being normalized by input scaling factors. The

two normalized FLC inputs are fuzzified using a fuzzy set of

five equally distributed, with 50% overlapping, Gaussian

membership functions. The fuzzy output, results from the fired

fuzzy rules of the FLC, changes to crisp values using the centre

of area defuzzification method. The output is defuzzified using

five equally distributed Gaussian membership functions. The

output of each controller, after being scaled by output scaling

factors, is added with a constant reference pulse width. The

values of constant pulse width together with the input/output

scaling factors are determined heuristically as G1=G10=

0.0034, G2=G11=0.007, G3=G12=70, G4=G7=0.002,

Fig. 2. The control block diagram of scenario I

Fig. 1. Stimulation phases according to knee position

G5=G8= 0.0067, G6=G9= -45, R1=R4=60, R2=70, R3=50.

The performance of this control approach can be evaluated

from Figs 3-5. It’s clear from Figs 3-4 that each muscle is

stimulated twice per cycle; once for push and once for resist

phase. Also both muscles received approximately equal amount

of stimulation and produced the same torque in successive

cycles which is a good indicator of a coordinated pedaling

movement.

From Fig. 5 it is clear that the control strategy was successful

in producing cycling movement. However, the dead points

resulted in rapid changes in the crank cadence.

B. FES cycling with a flywheel and electrical clutch

mechanism (Scenario II)

In this scenario it is aimed to perform FES cycling by

stimulating the quadriceps only with the help of a new assist

mechanism represented by a flywheel and an electrical clutch.

The flywheel, as energy storage device, is used to assist and

retard the cycling when necessary and to replace muscle

stimulation in the resist phase. The electrical clutch engages the

flywheel with the crank to absorb the excessive energy

produced by the leg and to store it as kinetic energy in the

flywheel. Also the flywheel, loaded with energy, is engaged by

the clutch to discharge its energy into the crank and support the

cycling in case the energy of the leg is not enough to pedal.

In this scenario, two fuzzy logic controllers are used to

control the amount of stimulation during the pushing phase,

while the resist phase is replaced by the assist mechanism. The

control block diagram can be seen in Fig. 6. The input/output

scaling factors and the reference pulse width used in this

scenario are the same as these of scenario I.

The flywheel engagement and disengagement is

implemented in a closed-loop approach as shown in Fig. 7. The

engagement logic is constructed depending on the

instantaneous values of the crank angular velocity and the

angular velocity of the flywheel during the cycling. If the

angular velocity of the crank is greater than the required

cadence, i.e. the crank is fast and excessive energy is available

in the system, and if the angular velocity of the flywheel is less

than that of the crank, i.e. the flywheel is ready to resist, the

clutch will engage the flywheel with the crank to absorb the

energy and slow down the movement. Also, if the angular

velocity of the crank is less than the required cadence, i.e.

needs assistance, and the angular velocity of the flywheel is

greater than that of the crank, i.e. the flywheel is ready to

assist, the clutch will engage the flywheel to discharge its

energy into the crank and speed up the system.

Fig. 6. The control block diagram of scenario II

Fig. 5. The angular velocity of the crank

Fig. 4. Muscle torque of left leg

Fig. 3. Muscle torque of right leg

The performance of this control strategy can be seen from

Figs 8-14. Figs 8-9 show the engagement and disengagement

of the flywheel, by the clutch, for assist and resist periods.

From Fig. 10 and Fig. 11 it is clear that each leg was stimulated

once per cycle while there was an increase in the torque

produced by the muscles which indicates an increase in the

work done by the muscle. It can be seen from Fig. 12 that the

energy absorption and energy release of the flywheel during

resist and assist periods respectively and the corresponding

effect on the angular velocity of the crank, as in Fig. 13. It is

clear from Figs 12-13 that the flywheel mechanism was

successful in reducing the crank speed when a resist action was

required and also was successful in speeding up the crank when

an assist action was required.

Fig. 14 shows a comparison between the cadence error in

both scenarios. It is clear that the flywheel mechanism was

effective in reducing the error in the crank cadence as well as

reducing the rapid change in the angular velocity at the cycling

dead points and consequently produced smoother cycling

Fig. 13. The angular velocity of the crank

Fig. 12. The angular velocity of the flywheel

Fig. 11. Muscle torque of right leg

Fig. 10. Muscle torque of left leg

Fig. 9. Flywheel engagement (resist period)

Fig. 8. Flywheel engagement (assist period)

Fig. 7. The flywheel engagement mechanism

cadence. From Fig. 15 it can be seen that the new assist

mechanism reduced the overall stimulation (from 600µs to

338µs) on the muscles. Reducing the stimulation intensity on

the muscle delays the appearance of muscle fatigue, and hence

the new mechanism promotes extended FES cycling exercise.

IV. CONCLUSION

Coordinated FES cycling can be performed by stimulating

single muscle group, the quadriceps only. Using a closed-loop

control approach, a flywheel and electrical clutch mechanism

can be utilized to assist the legs and improve the cycling

performance. The new assist mechanism has shown superior

performance in terms of reducing the cadence error and the

sharp change in speed at the cycling dead points which results

in smoother and more comfortable cycling. The introduced

control approach with the new assist mechanism has reduced

the overall stimulation intensity on leg muscles by

approximately 43% which promotes delayed muscle fatigue

and prolonged FES cycling exercise.

REFERENCES

[1] G. M. Davis, N. A. Hamzaid, and C. Fornusek, Cardiorespiratory, metabolic, and biomechanical responses during functional electrical stimulation leg exercise: health and fitness benefits, Artificial Organs, vol. 32 (2008) pp. 625-629.

[2] L. Griffin, M. Decker, J. Hwang, B. Wang, K. Kitchen, Z. Ding, and J. Ivy, Functional electrical stimulation cycling improves body composition, metabolic and neural factors in persons with spinal cord injury, Journal of Electromyography and Kinesiology, vol. 19 (2009) pp. 614-622.

[3] S. P. Hooker, A. E. Scremin, D. L. Mutton, C. F. Kunkel, and G. Cagle, Peak and submaximal physiologic responses following electrical stimulation leg cycle ergometer training, Journal of rehabilitation research and development, vol. 32 (1995) pp. 361-366.

[4] A. Frotzler, S. Coupaud, C. Perret, T. H. Kakebeeke, K. J. Hunt, N. N. Donaldson, and P. Eser, High-volume FES-cycling partially reverses bone loss in people with chronic spinal cord injury, Bone, vol. 43 (2008) pp. 169-176.

[5] M. L. Sipski, J. A. Delisa, and S. Schweer, Functional electrical stimulation bicycle ergometry: patient perceptions, American Journal of Physical Medicine & Rehabilitation, vol. 68 (1989) pp. 147-149.

[6] K. Chen, S. C. Chen, K. H. Tsai, J. J. J. Chen, N. Y. Yu, and M. H. Huang, An improved design of home cycling system via functional electrical stimulation for paraplegics, International Journal of Industrial Ergonomics, vol. 34 (2004) pp. 223-235.

[7] D. Pons, C. Vaughan, and G. Jaros, Cycling device powered by the electrically stimulated muscles of paraplegics, Medical and Biological Engineering and Computing, vol. 27 (1989) pp. 1-7.

[8] M. Gföhler, M. Loicht, and P. Lugner, Exercise tricycle for paraplegics, Medical and Biological Engineering and Computing, vol. 36 (1998) pp. 118-121.

[9] J. Petrofsky and J. Smith, Three-wheel cycle ergometer for use by men and women with paralysis, Medical and Biological Engineering and Computing, vol. 30 (1992) pp. 364-369.

[10] J. Chen, N. Y. Yu, D. G. Huang, B. T. Ann, and G. C. Chang, Applying fuzzy logic to control cycling movement induced by functional electrical stimulation, IEEE Transactions on Rehabilitation Engineering, vol. 5 (1997) pp. 158-169.

[11] J. J. Abbas and R. J. Triolo, Experimental evaluation of an adaptive feedforward controller for use in functional neuromuscular stimulation systems, IEEE Transactions on Rehabilitation Engineering, vol. 5 (1997) pp. 12-22.

[12] J. Riess and J. J. Abbas, Adaptive neural network control of cyclic movements using functional neuromuscular stimulation, IEEE Transactions on Rehabilitation Engineering, vol. 8 (2000) pp. 42-52.

[13] C. S. Kim, G. M. Eom, K. Hase, G. Khang, G. R. Tack, J. H. Yi, and J. H. Jun, Stimulation pattern-free control of FES cycling: simulation study, IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews, vol. 38 (2008) pp. 125-134.

[14] K. J. Hunt, B. Stone, N. O. Negard, T. Schauer, M. H. Fraser, A. J. Cathcart, C. Ferrario, S. A. Ward, and S. Grant, "Control strategies for integration of electric motor assist and functional electrical stimulation in paraplegic cycling: utility for exercise testing and mobile cycling," Neural Systems and Rehabilitation Engineering, IEEE Transactions on, vol. 12, pp. 89-101, 2004.

[15] T. Takahashi, M. Takazawa, Y. Nishiyama, E. Nakano, and Y. Handa, FES cycling chair for the lower limbs disabled people with electric motor power assist, in 9th Annual Conference of the International FES Society, September (2004).

[16] R. Massoud, Intelligent control techniques for spring-assisted FES-cycling, Ph.D. thesis, Dept. Auto. Cont. and Sys. Eng., Univ. of Sheffield, Sheffield, UK, 2007.

[17] S. C. Abdulla, O. Sayidmarie, S. Gharooni, and O. Tokhi, Modeling and control of a novel FES driven assisted cycling mechanism, 17’th International Conference on Methods and Models in Automation and Robotics, Poland (2012) pp. 464-469.

[18] D. A. Winter, Biomechanics and motor control of human movement, second edition, Wiley-Interscience, New York, 1990.

[19] F. Zajac, E. Topp, and P. Stevenson, A dimensionless musculotendon model, Proceedings IEEE Engineering in Medicine and Biology, (1986) pp. 26-31.

[20] H. Makssoud, D. Guiraud, and P. Poignet, Mathematical muscle model for functional electrical stimulation control strategies, IEEE International Conference on Robotics and Automation, New Orleans, Vol. 2 (2004) pp. 1282-1287.

[21] M. Ferrarin and A. Pedotti, The relationship between electrical stimulus and joint torque: a dynamic model, IEEE Transactions on Rehabilitation Engineering, vol. 8 (2000) pp. 342-352.

Fig. 15. Integral of stimulation intensity in both scenarios

Fig. 14. Cadence error in both scenarios