Human-based Lower Limb Movement Assistance and Rehabilitation

through an Actuated Orthosis

Samer Mohammed, Hala Rifai, Walid Hassani and Yacine Amirat

LISSI Lab, University of Paris Est Cr´eteil-(UPEC), 122 Rue Paul Armangot, Vitry Sur Seine, France

Keywords:

Wearable Robots, Movement Assistance, Rehabilitation Robotics.

Abstract:

In this paper, an overview of the project EICOSI (Exoskeleton Intelligently COmmunicating and Sensitive to

Intention) conducted at the LISSI/UPEC Lab will be presented. This work aims to control a knee joint actuated

orthosis while tracking a desired trajectory or following the wearer’s intention. The proposed control strategies

ensure satisfactory performances in terms of trajectory tracking, intention detection and torque generation

during rehabilitation tasks and assistive movements of the wearer’s lower limbs.

1 INTRODUCTION

The last decade has shown great advances in the de-

velopment of robots that interact closely with human.

The sensor and actuator technologies have also shown

great evolution particularly in terms of energy, porta-

bility and miniaturization. The related applications

concern chieﬂy the assistance and rehabilitation of de-

pendent/elderly people. In 2000, more than 60 million

people in Europe were aged 65 or over. The rise in

life expectancy,set to continue, combined with falling

birth rates, will accelerate the ageing of the popula-

tion (Mohammed et al., 2012). This will certainly

have a great impact on the development of assisting

robotic devices. An exoskeleton is a mechanical ac-

tuated structure with several degrees of freedom en-

abling the assistance or rehabilitation at multiple joint

levels such as the robot suit HAL (Tsukahara et al.,

2010). Particularly, the knee orthoses have been de-

veloped to tackle musculoskeletal impairments at the

knee joint level. The main pointed purposes are the

rehabilitation and the assistance. To ensure the ﬁrst

purpose, passive or active rehabilitation can be en-

sured. A patient-directed orthosis is used after to-

tal knee arthroplasty for knee stiffness reduction and

range of motion increase by executing static progres-

sive stretches in (Bonutti et al., 2010). The reha-

bilitation has also been addressed in (Schmitt et al.,

2004) using the hybrid orthosis cyberthosis, activated

by a functional electrical stimulation of the muscles

and controlled using a proportional, integral, deriva-

tive (PID) controller in order to train the knee joint

and the muscles actuating it, for ﬂexion/extension

movements. On the other hand, the TUPLEE ortho-

sis (Fleischer and Hommel, 2008) aims to assist the

wearer during various activities like standing, sitting

and climbing. The user’s intention is determined us-

ing Electromyogram (EMG) electrodes placed at the

thigh and the control torque behaves as an ampliﬁer

of the knee torque. An elastic knee brace has been

designed in (Cherry et al., 2006) to add a parallel

stiffness to the knee allowing to compensate the in-

crease of the knee joint’s stiffness during running.

Impedance control has been addressed in (Aguirre-

Ollinger et al., 2007) to change the damping parame-

ter and in (Aguirre-Ollinger et al., 2010) to change the

inertia in order facilitate the ﬂexion/extension move-

ments.

The present paper deals with the control of a lower

limb orthosis applied at the knee joint level. Knee

joint has a great importance in maintaining the hu-

man stability during the different daily living activ-

ities. Two case studies are proposed: the ﬁrst one

deals with the passive rehabilitation process. A de-

sired knee joint angle is needed and is supposed de-

livered by the rehabilitation doctor. The wearer’s con-

tribution is considered active if it acts in the desired

position direction and is considered as external per-

turbation if it acts in the opposite direction. In this

case the priority is given to the desired task comple-

tion independently from the human contribution (Ri-

fai et al., 2011). The second case study concerns

the human based control and lies within the assistive

strategy. Muscular activities of the muscles spanning

the knee joint are measured and the resulting knee

joint torque is ampliﬁed through the use of the or-

205

Mohammed S., Rifai H., Hassani W. and Amirat Y..

Human-based Lower Limb Movement Assistance and Rehabilitation through an Actuated Orthosis.

DOI: 10.5220/0004664902050209

In Proceedings of the International Congress on Neurotechnology, Electronics and Informatics (RoboAssist-2013), pages 205-209

ISBN: 978-989-8565-80-8

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

thosis to ensure a desired movement initiated by the

wearer (Hassani et al., 2013). Asymptotic stability

of the knee joint orthosis is guaranteed in both cases.

Control law should take into consideration constraints

related to the safety of the mechanism since it is in di-

rect relation with the human body. In both case stud-

ies, the effectiveness of the proposed control strategy

is tested in real-time with a healthy subject.

The rest of the paper is organized as follows: Section

2 shows the system modeling and parameters identi-

ﬁcation. Section 3 presents the passive rehabilitation

control strategy while section 4 presents the subject

centered control strategy case study. Section 5 shows

the results in both cases and section 6 concludes the

paper.

2 SHANK-ORTHOSIS

MODELING AND

PARAMETERS

IDENTIFICATION

In the project EICOSI, the orthosis has a simple de-

sign, easy to don and doff, and match the lower limb

geometry which makes it very practical to use by el-

derly and dependent people. The orthosis is one struc-

ture having two segments related along a rotational

axis. The ﬁrst segment embodies the thigh while

the second one embodies the shank and are ﬁxed to

the wearer by means of straps (Figure 1). The or-

thosis and the human leg have, then, the same ro-

tational degree of freedom at the knee joint level.

The shank/orthosis parameters are identiﬁed using the

weighted least square optimization method. The mass

of the shank and the position of its center of grav-

ity are determined based on (Winter, 2009) given the

height and weight of the subject. The other param-

eters are identiﬁed using the passive pendulum test.

To identify the orthosis’ parameters, an excitation se-

quence describing the trajectory of the angle is prede-

Figure 1: EICOSI knee-joint orthosis.

ﬁned as well as the angular velocity and acceleration

(Swevers et al., 1997). The torque developed by the

actuator during the tracking of the excitation trajec-

tory is computed using a current sensor.

3 PASSIVE REHABILITATION

CONTROL STRATEGY

The passive rehabilitation is performed after surgi-

cal operations of articular tissues as well as rehabil-

itation of spinal cord injured patients (Jansen et al.,

1996) (Rudhe et al., 2012) (Chang et al., 2013). In

this case study, the wearer is considered either com-

pletely passive, therefore, the wearer does not develop

any human control torque or paritally passive where

the wearer is developing an assistive but not sufﬁ-

cient torque. If the wearer develops resistive torque,

this movementwill be considered as perturbing move-

ment and will be rejected by the controller. In this

case study, the control law is based on a gravity com-

pensation and ensures the tracking of a desired trajec-

tory. The proposed strategy ensures the asymptotic

stability of the orthosis/wearer for ﬂexion-extension

movements. The stance phase of locomotion has not

been taken into account. The control is guaranteed

to be bounded in order to take into account limited

power support. Also the saturation of the actuator is

used explicitly in the control law in order to prevent

problems related to nonlinearities of the shank-foot

orthosis. Quick friction variations induced by unpre-

dictable movementsmay cause unacceptable behavior

of the orthosis, also high value of the control torque

necessitates high power to ensure it, which cannot

be achieved in wearable robots because it affects the

safety of the wearer. The saturation based control law

can take into account these limitations. Consequently,

the actuator magnitude limitation is taken into con-

sideration in the design of the control torque in or-

der to avoid irreversible damages. The proposed con-

trol law is robust with respect to external perturba-

tions. Indeed, a misstep can be caused by a wrong

movement at the knee level, it can cause instability

or even falling down. Therefore, one main property

of the control law is to regain the intended position

whenever an unpredictable ﬂexion occurs. Experi-

ments that consist on blocking the shank for a short

period by an external operator have been conducted

and shown satisfactory results in terms of quickly re-

gaining the desired position. The magnitude of the re-

sulted perturbation remains within the saturation lim-

its imposed by the controller.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

206

4 SUBJECT CENTERED

CONTROL STRATEGY

An ”Assistance as Needed” approach is developed to

allow a ﬂexible and smooth orthosis/wearer interac-

tion. In this case study, a realistic bio-inspired muscu-

loskeletal model is used to control the orthosis move-

ments (Hassani et al., 2013) . Parameters such as mus-

cle length variations, activation and contraction dy-

namics and moment arms are taken into account. This

will allow a better estimation of the wearer’s inten-

tion. The identiﬁcation of the above parameters was

made through an unconstrained optimization problem

formulation. Secondly, a control law strategy is de-

veloped to guarantee asymptotic stability of the knee

joint orthosis as well as ﬂexible interaction between

human and exoskeleton. This control law has also

shown its robustness with respect to external distur-

bances. In this case study, the wearer is supposed to

developing a torque to move his/her leg. The orthosis

is assisting the wearer by delivering the complemen-

tary torque, necessary to achieve a given desired task.

The subject’s intention is taken into account through

the human developed torque while the controller’s pa-

rameters are adapted automatically with respect to

the wearer’s changing human contribution during the

completion of the task. This approach can be used

particularly during a rehabilitation process as the con-

tribution of the orthosis torque decreases with the im-

provement in the performance of the wearer. Indeed,

the controller continuously attempt to reduce its as-

sistance ratio when the wearer develops an effort to

accomplish a given task. On the other side, when the

user moves largely away of the desired joint trajec-

tory, the controller develops a counter-torque to bring

the wearer limb back to the desired trajectory or to its

vicinity.

5 RESULTS

The control law is tested in real-time using the

EICOSI orthosis of the Laboratory of Images, Sig-

nals and Intelligent Systems (LISSI) of the University

of Paris-Est Crteil (UPEC). The mechanical structure

of the orthosis consists of two segments attached to

the thigh and shank respectively by means of straps,

with a rotation axis at the knee level. The orthosis is

actuated using a brushless DC motor (BLDC) chosen

because it delivers a relatively high torque and runs

smoothly at low speeds. The maximal torque that can

be delivered by the actuator is equal to 25 N.m. The

orthosis is also equipped with an incremental encoder

that delivers the angle of the shank segment relative to

the thigh one. The control torque is computed using

a controller board (dSPACE-DS1103) equipped with

an IBM processor running at 400Mhz. The controller

takes the measurement of the angle delivered by the

EICOSI’s sensor and the angular velocity obtained by

a simple derivation as well as the desired angle and

velocity. The controller board delivers the pulse width

modulation (PWM) level to control the actuator’s ve-

locity. The control loop runs at 1 kHz, ﬁxed due to

current and position sensors constraints. The exper-

iments are conducted on a healthy subject having 35

years old, weighing 94Kg and measuring 1.82m.

0 5 10 15 20 25 30 35

−1.5

−1

−0.5

0 5 10 15 20 25 30 35

−2

0 5 10 15 20 25 30 35

−10

0 5 10 15 20 25 30 35

−100

100

0 5 10 15 20 25 30 35

−100

100

θ(rad)

θ(rad/s)

τ(N · m)RF (µV)BF (µV)

time (s)

Figure 2: Sine reference trajectory with assisting human

effort in the time interval [8.5, 28.5] s: The ﬁrst two plots

represent a good tracking between the current and desired

knee-joint angles and angular velocities, the current values

are plotted with continuous blue line and the desired ones

with red dashed lines. The control torque τ has lower val-

ues during the assistance phase detected by the RF and BF

muscles activities.

In ﬁgure 2, the subject is able to move his leg but

does not have sufﬁcient power to achievea movement.

In other words, the subject deliversa part of the torque

necessary for the movement and the exoskeleton is

supposed to help him ensuring the desired movement

by delivering the complementary torque. This strat-

egy is known as the assistance-as-needed. The subject

is delivering an effort in the time interval [8.5, 28.5]s

that goes in the same direction as the desired trajec-

tory. It can be seen that the Rectus Femoris (RF) mus-

cle is active during the extension (sine with positive

slope) while the Biceps Femoris (BF) during the ﬂex-

ion (sine with negative slope) in the last two plots of

Figure 2. However, this effort is not sufﬁcient to en-

Human-basedLowerLimbMovementAssistanceandRehabilitationthroughanActuatedOrthosis

207

sure the tracking of the desired trajectory. The control

is therefore only assisting the subject during this inter-

val. One can notice that the magnitude of the control

torque is lower than that delivered outside the assis-

tance phase, i.e. in the time intervals [0,8.5[s and

]28.5, 38] s. The control law ensures a good adap-

tation and trajectory tracking (ﬁrst two plots of Fig-

ure 2) during the assistance and non-assistance phases

which guarantees the safety of the wearer.

Figure 3: Assistance results (A): Desired position (dashed

blue) and current position (red), (B): Human torque (dashed

blue) and exoskeleton torque (red).

In ﬁgure 3, the control is based on the wearer’s

intention. The wearer’s torque is estimated using the

lower limb anatomical model based on the use of the

modiﬁed hill-type muscle model. The raw EMG sig-

nals are measured from the quadriceps and hamstrings

muscles. During this experiment, the wearer is asked

to produce free ﬂexion / extension movements of the

knee joint. The advantage of the proposed approach

consists in its nature which is subject centered where

no predeﬁned position trajectory is imposed to the

wearer.

6 CONCLUSIONS

This paper treated the control of a knee-joint ortho-

sis. A model, of the shank and orthosis, is proposed

and its parameters has been identiﬁed. Two control

strategies were proposed, the ﬁrst one deals with the

passive rehabilitation process. A desired knee joint

angle is needed and is supposed delivered by the re-

habilitation doctor. In this case the priority is given

to the desired task completion independently from

the human contribution. The second case study con-

cerns the human based control and lies within the as-

sistive strategy. Muscular activities of the muscles

spanning the knee joint are measured and the result-

ing knee joint torque is ampliﬁed through the use of

the orthosis to ensure a desired movement initiated

by the wearer. Experimental validations of the pro-

posed strategies were conducted in real-time with a

young healthy subject using the EICOSI orthosis of

the LISSI Lab.

ACKNOWLEDGEMENTS

The EICOSI project is sponsored by the regional

council of Ile-De-France.

REFERENCES

Aguirre-Ollinger, G., Colgate, J., Peshkin, M., and

Groswami, A. (2007). A 1-DOF assistive exoskeleton

with virtual negative damping: Effects on the kine-

matic response of the lower limbs. In Proceedings

of International Conference on Intelligent Robots and

Systems, pages 1938–1944, San Diego, CA, USA.

Aguirre-Ollinger, G., Colgate, J., Peshkin, M., and

Groswami, A. (2010). Design of an active 1-DOF

lower-limb exoskeleton with inertia compensation. In-

ternational Journal of Robotics Research, 30(4):486–

499.

Bonutti, P., Marulanda, G., McGrath, M., Mont, M., and

Zywiel, M. (2010). Static progressive stretch im-

proves range of motion in arthroﬁbrosis following toal

knee arthroplasty. Knee surgery, sports traumatology,

arthroscopy, 18(2):194–199.

Chang, Y., Liang, J., Hsu, M., Lien, H. Fang, C., and

Lin, C. (2013). Effects of continuous passive mo-

tion on reversing the adapted spinal circuit in humans

with chronic spinal cord injury. Archives of Physical

Medicine and Rehabilitation.

Cherry, M.-S., Choi, D.-J., Deng, K.-J., Kota, S., and Fer-

ris, D.-P. (2006). Design and fabrication of an elastic

knee orthosis: Preliminary results. In Int. design engi-

neering technical conf. & Conputers and information

in engineering conf., pages 565–573, USA.

Fleischer, C. and Hommel, G. (2008). A human-

exeskeleton interface utilizing electromyography.

IEEE Transactions on Robotics, 24(4):872–882.

Hassani, W., Mohammed, S., and Amirat, Y. (2013). Real-

time emg driven lower limb actuated orthosis for as-

sistance as needed movement strategy. In Robotics:

Science and Systems Conference.

Jansen, C., Windau, J., Bonutti, P., and Brillhart, M. (1996).

Treatment of a knee contracture using a knee orthosis

incorporating stress-relaxation techniques. Journal of

Physical Therapy Association, 76(2):182–186.

Mohammed, S., Amirat, Y., and Rifai, H. (2012). Lower-

limb movement assistance through wearable robots:

state of the art and challenges. Advanced Robotics,

26(1-2):1–22.

Rifai, H., Hassani, W., Mohammed, S., and Amirat, Y.

(2011). Bounded control of an actuated lower limb or-

thosis. In Decision and Control and European Control

Conference (CDC-ECC), 2011 50th IEEE Conference

on, pages 873–878.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

208

Rudhe, C., Albisser, U., Starkey, M. Curt, A., and Bol-

liger, M. (2012). Reliability of movement workspace

measurements in a passive arm orthosis used in spinal

cord injury rehabilitation. Journal of Neuroengineer-

ing and Rehabilitation, 9(37).

Schmitt, C., Metrailler, P., Al-Khodairy, A., Brodard, R.,

Fournier, J., Bouri, M., and Clavel, R. (2004). A

study of a knee extension controlled by a closed loop

functional electrical stimulation. In 9

annual confer-

ence of the International FES society, pages 135–137,

Bournemouth.

Swevers, J., Ganseman, C., Tukel, D. B., De Schutter, J.,

and Van Brussel, H. (1997). Optimal robot excitation

and identiﬁcation. Robotics and Automation, IEEE

Transactions on, 13(5):730–740.

Tsukahara, A., Kawanishi, R., Hasegawa, Y., and Sankai, Y.

(2010). Sit-to-stand and stand-to-sit transfer support

for complete paraplegic patients with robot suit HAL.

Journal of Advanced Robotics, 24(1):1615–1638.

Winter, D. A. (2009). Biomechanics and motor control of

human movement. Wiley.

Human-basedLowerLimbMovementAssistanceandRehabilitationthroughanActuatedOrthosis

209