design considerations of a lower limb exoskeleton system...

Applied Bionics and Biomechanics 11 (2014) 119–134DOI 10.3233/ABB-140099IOS Press

119

Design considerations of a lower limbexoskeleton system to assist walking andload-carrying of infantry soldiers

Seungnam Yua,∗, Changsoo Hanb and Ilje Choa

aRemote System Technology Development Section, Korea Atomic Energy Research Institute, Deajeon, South KoreabDepartment of Robot Engineering, Hanyang University, Ansan, South Korea

Abstract. This paper describes the development of a wearable exoskeleton system for the lower extremities of infantry soldiersand proposes appropriate design criteria based on existing case studies. Because infantry soldiers carry a variety of equipment,the interference with existing equipment and additional burden of the exoskeleton support system should be minimized. Recentstudies have shown that a user only needs to be supported in the gravitational direction when walking on flat terrain; however, activejoints are necessary to support walking over rough and sloped terrain such as mountains. Thus, an underactuated exoskeletonsystem was considered: passive joints are applied to the hip and ankle joints, and active joints are applied to the knee joints toexploit the dynamic coupling effect of the link structure and muscular activation patterns when the user is going up and downstairs. A prototype of the exoskeleton system was developed and validated through a simple stair-climbing experiment.

Keywords: Exoskeleton, wearable robot, infantry soldier, underactuated system, weight suppot mechanism (WSM)

1. Introduction

According to records from the past 10 years, heavypersonal belongings on soldiers have increased [1,2]. This inevitably increases the possibility of mus-culoskeletal diseases; likewise, injuries to soldiersin circumstances other than battles, i.e., nonbattleinjuries, have emerged as problems that critically affectthe physical condition of soldiers [3–6]. The increasein these heavy personal belongings results from indi-vidual soldiers’ requests to be sufficiently equippedto ensure personal safety in any circumstance as wellas the increase in additional equipment used duringfield operations. For example, the maximum weight ofpersonal belongings recommended for a U.S. infantrysoldier is known to be approximately 23 kgf, whereas

∗Corresponding author: Seungnam Yu, Remote System Technol-ogy Development Section, Korea Atomic Energy Research Institute,Deajeon, South Korea. E-mail: [email protected].

the practical weight of personal belongings is knownto reach no less than 45–60 kgf [1]. This tendency istrue for other countries’ infantry systems as well; thus,it is necessary to improve transportation equipmentfor heavy personal belongings, to minimize nonbat-tle injuries such as musculoskeletal diseases and toenhance the fighting power and mobility of infantrysoldiers. In particular, in the field of national defense,many wearable lower extremity exoskeleton systemshave been proposed, focusing on relieving the weightof the backpack, which is one of the major heavy per-sonal belongings of individual wearers [7–11]; someof these systems emphasize their effectiveness by mea-suring the metabolic rate changes of wearers [12, 13].However, as proposed in some studies, these systemstend to hinder the natural gaits of wearers owing tothe excessive application of actuators or the kinematicproblems of exoskeletons, and this tendency is a pri-mary cause of the additional metabolic consumptionof wearers [14]. Considering the nature of infantry

1176-2322/14/$27.50 © 2014 – IOS Press and the authors. All rights reserved

mailto:[email protected]

120 S. Yu et al. / Design considerations of a lower limb exoskeleton system

duties in open fields, natural walking and runningare essential elements; thus, the exoskeleton systemthese soldiers wear should be designed to avoid hin-dering their natural walking and to relieve the weightof heavy personal belongings. This study analyzes andproposes various design requirements through exist-ing research and development cases for implementingthe wearable exoskeleton system for infantry soldiers.Furthermore, a muscular power assistance system foroffsetting muscular stress is proposed to attribute to theweight of personal belongings. This study focuses onthe weight-supporting mechanism (WSM) for majorheavy personal belongings such as backpacks whileaddressing a general wearable lower-extremity muscu-lar power assistance system. The conclusion proposesrequirements for improving personal portable equip-ment related to the exoskeleton system for infantrysoldiers.

2. Consideration of the motional characteristicsof human bodies and proposal of theexoskeleton joint structure

2.1. Outline

Because the infantry is a branch of the military ser-vice in which soldiers primarily conduct their duties inopen fields, superior strength and exercise capacitiesare required. In particular, regarding exercise capacity,various motions such as walking and running erect,sitting on the knees, and crawling are required accord-ing to behavior types such as shooting, moving, andequipment handling. It is critical that the exoskele-ton system not restrict these behaviors. Although thedetailed structure will be discussed in Section 3, theexoskeleton system, when various motions are per-formed as shown in Fig. 1, should not hinder themotions of wearers. Consequently, structurally specificparts should not excessively protrude nor weigh heav-ily, and the system should be easily worn and removed.

2.2. Two-jointed muscle

Prior to determining the actuating structure of theexoskeleton system, the characteristics of the humanbody’s muscles should be considered. Generally, flex-ors and extensors belonging to the same joint canalso function as the flexors and extensors of the adja-cent joint. These muscles are designated two-jointedmuscles (TJMs) and are sometimes called biarticu-

late muscles or biaxial muscles (Fig. 2). These TJMs,which are formed over two joints, are distributed overshoulders, elbows, hip joints, knee joints, and anklejoints and enable human bodies to generate various andcomplicated motions. Among these joints, this studyconcentrates on hip joints. Joints such as semitendi-nosus, semimembranosus, and biceps femoris (longhead) are simultaneously involved in hip joint exten-sion and knee flexion. In other words, the simpleact of raising a straightened leg requires the interac-tion of various muscles between hip joints and kneejoints. Thus, the simple application of actuators to hipjoints/knee joints/ankle joints for the exoskeleton sys-tem, which is operated in complicated open fields, maypossibly hinder the wearers’ motions and movements.

In particular, active exoskeleton joints applyingactuators require synchronizing signals from the wear-ers for actuation. However, in field operations, sensorsreceiving these synchronizing signals, irrespectiveof the human body’s contact methods (i.e., elec-tromyogram sensors or muscle stiffness sensors) orindirect/noncontact methods (i.e., force sensors, pres-sure sensors attachable to shoes, acceleration sensorsfor each link, encoder for each joint [19, 20]), can-not infer all of the wearers’ intentions for complicatedmotions. In particular, in the case of electromyogramsensors, most studies have examined using electricsignals individually extracted from each muscle; aspreviously stated, because TJMs play an important rolein the maneuvers performed by an infantry soldier, ifelectromyogram signals are used, signals from vari-ous muscles should be comprehensively analyzed forpractical use.

2.3. Joint composition of the lower extremityexoskeleton system

Of the various lower extremity exoskeleton systemsintroduced in recent years, the BLEEX system of U.C.Berkeley, HULC of Lockheed Martin, and EXO seriesof Sarcos are widely known military exoskeleton sys-tems for infantry soldiers [7–10]. Further studies havebeen continuously conducted on these systems, andthe overall direction of improvement can be defined asthe simplification of the exoskeleton and actuator sys-tem. In other words, the current independent sourcetechnology lacks technical maturity for applying anexoskeleton system using a number of high mobilityactuators for individual soldiers; in particular, it is dif-ficult to persuade soldiers who wear various personal

S. Yu et al. / Design considerations of a lower limb exoskeleton system 121

Fig. 1. Field operations of infantrymen in carrying loads [15–18].

portable equipment to wear an additional heavyexoskeleton system, even when considering advan-tages such as supporting the weight of heavy personalbelongings. Considering these overall limitations, theactuating method for the exoskeleton system used inopen fields is proposed as follows.

• As exoskeleton joints corresponding to jointsdirectly related to TJMs (hip joints and anklejoints), passive, or quasi-passive joints can beapplied; for the application of active joints, theactuating mechanism of actuators needs to havecompliance or backdrivability.

• The overall exoskeleton system should distributethe weight of the backpack to body parts belowthe waist and simultaneously support the weight tominimize the weight burden on the wearer. At thistime, the mechanism for weight support shouldsimultaneously function as passive hip joints.

Based on these considerations, this study proposesan underactuated system as an exoskeleton system forinfantry soldiers, implementing passive joints for hipjoints to enable walking and the WSM for backpacks,active joints using an electric motor for knee joints, andpassive joints using an elastic body for ankle joints. For


Fig. 2. Two-jointed muscle for hip joint.

knee joints, rotating and linear actuators were variouslyapplied based on existing studies, and cases in whichrotating active joints were used (i.e., previous stud-ies) are given as references [21]. Sensing of wearers’motion intentions for actuating active joints will not bediscussed in this study. However, since it is difficult tomeasure bio-signals using an electromyogram sensorbecause infantry soldiers operate in open fields, phys-ical sensors such as pressure, and acceleration sensorsare considered to be appropriate.

2.4. Intersegmental dynamics

Infantry soldiers frequently traverse slopes such asmountains to conduct operations. The passive jointexoskeleton system [22] for walking on flat landand supporting the weight of backpacks, which wasproposed in a recent similar study, cannot assist mus-cular power when moving on slopes; this is a resultof focusing on lightening and supporting the weightof the system. In this respect, systems such as theunderactuated system proposed in this study can beused on sloped terrains by supporting the wearers’muscular power and the weight of backpacks and

ensuring a moderate wearing sensation. In terms ofsystem efficiency, the placement of the optimized actu-ators is particularly important, which implies that, aspreviously stated, the placement of actuators on theexoskeleton (passive hip joints, active knee joints, andpassive ankle joints) needs to be considered in termsof muscular activation characteristics and dynamics.According to existing studies on the body’s muscu-lar functions related to walking, the motion of liftingheels when overcoming an obstacle or moving up stairsconsists of bending hip joints, knee joints, and anklejoints. At this time, the energy transferred to each jointis appropriately adjusted according to the heights ofobstacles or stairs. The motions of bending hip andankle joints are similar to passive motions, which arecaused more by the active force of these joints thanby the motion of rotating knee joints. This segmenta-tion of joint moment can be analyzed as a dynamicaloptimization method based on the interaction betweensegments of the human nervous system to minimizeenergy consumption related to motions such as over-coming obstacles or moving up stairs [23]. This is alsoverified by the dynamical coupling effect in the under-actuated systems [24, 25]. In other words, it can be


inferred that a certain level of muscular power assis-tance with respect to hip joints can be achieved onlythrough the actuating power of the knee joints. In thisrespect, as proposed in existing studies, experimentalresults indicate that the overall muscular power assis-tance for major activated muscles of wearers’ lowerextremities with respect to the motion of going up stairs(Figs. 11 and 12) can be achieved only by actuating theexoskeleton knee joints [21].

3. Kinematic structure of an exoskeletonsystem for infantry soldiers

3.1. Leg structure

In this section, the structure related to the link shapeof the lower extremity exoskeleton system is consid-ered. If LThigh is defined as the length of the thigh andLShin as the length of the shin of the exoskeleton, theleg length of the exoskeleton is advantageous in thefollowing proportions.

LThigh

LShin

< 1 (1)

This is because the exoskeleton creates swings inthe wearer’s lower extremity on which the exoskeletonis attached; at this time, the overall trajectory of theseswings depends on the trajectory of the wearer’s feet. Inother words, when regarding the feet of the exoskele-ton as the end-effector of the manipulator, the thighand shin parts correspond to the 2-link manipulator;the swing of lower extremities during the swing phaseof general walking cycles shows the process from theinitial swing section to the landing point, similar to themotion of drawing a circular arc with a full-stretchedarm. At this time, in terms of manipulability, the struc-ture in Equation (1) can gather higher speed with thesame energy, and in terms of wearers, more naturalwalking can be implemented with less force. This isalso advantageous for the military purpose in termsof the efficiency of walking and stride control whileconnected to the exoskeleton [26]. Consequently, theB-type structure is more advantageous than the A-typestructure, as shown in existing study cases in Fig. 3.

3.2. Mechanism of ankle joints

The mechanism of ankle joints of the exoskeleton,in general cases, is an important linking point between

the wearer and the exoskeleton; additionally, it com-prehensively supports heavy personal belongings andthe weight of the exoskeleton itself. In previous stud-ies [21], by using the method mentioned in Section 2,passive ankle joints, which are highly elastic and com-posed of hyper carbon that can withstand the weight ofthe entire upper part, were applied. This enables wear-ers to perform an auxiliary function for natural forwardwalking by emitting energy, which is stored by theelastic deformation of the walking mechanism in theprocess of heel contact after the swing section, duringthe stance and toe-off sections. In recent studies, a sepa-rate actuator has not been used for ankle joints; in somesystems, the stiffness of ankle joints can be adjusted.It is known that ankle joints can adjust elasticity, andin particular, in the case of the underactuated system,they can also perform tuning functions such as adjust-ing the motion area of other adjacent passive joints[22]. Consequently, the ankle joints of the exoskeletoncan exhibit sufficient performance with the followingfunctions: 1) the linking point to wearers; 2) the transferof the weight of the upper exoskeleton and the externalheavy personal belongings to the ground; 3) the pro-vision of a stable wearing sensation by appropriatelylimiting the turning force or converting energy; and 4)the support of momentum.

3.3. WSM for backpacks

3.3.1. OutlineBackpacks have the heaviest weight among the

personal portable equipment of general soldiers. Thelower extremity exoskeleton also has the primary pur-pose of supporting the weight of backpacks. In thecase of this WSM, the weight support by applying theactive joints for hip joints can be considered; however,as previously stated, we propose a method to apply aWSM that can simultaneously perform the functionsof supplying passive joints for the hip joints of theexoskeleton and of supporting the weight of backpacks.Various WSMs for backpacks have been proposedsince the early 2000 s for national defense applica-tions. Recently, as introduced in the previous section,WSMs with a planar linkages have been proposed [21,22]. However, even these systems show differencesin function according to whether the active adjust-ment of stiffness is possible; in particular, the systemproposed by Hollander actively adjusts the stiffnessby using control signal patterns, which are calculatedin response to the weight of backpacks and walking


A B

Fig. 3. Comparison of linkage structures of exoskeleton systems [22, 27].

cycles of the wearers [22], and a separate external elec-trical power is installed for this purpose. In addition,the system proposed by Yu was designed to manuallyadjust the stiffness of springs according to the weightof wearers’ heavy personal belongings and their indi-vidual preferences for motion range while walking;thus, a separate external electrical power is not required(Of course, this kind of system used an external powerto operate the actuator of an active joint.). In partic-ular, in the latter case, the stiffness was selected andapplied by the constant force mechanism (WSM) ana-lytical method based on springs; by using this method,the mechanism itself was maintained in a constantlydeformed posture in response to the external weight(the weight of backpacks) [21] (Fig. 4).

Generally, the frequency of jogging is approxi-mately 3 Hz per s [28]. If this applies to soldiers infull gear, backpacks vibrate approximately 3 times persecond while running. On the other hand, the verticalground reaction force is known to be approximately 2.5times greater than their weight; if this applies to back-pack wearers, the impulsive-weight approximately 2.5times greater than the weight of the backpack is loadedon their shoulders when they land on the ground whilewalking [28]. An existing analytic study of carry-ing poles indicated that a considerable amount of the

weight loaded on the shoulders can be reduced bycomparing the vibration frequency of heavy personalbelongings, which are supported by the shoulders, withthe human body’s vertical frequency while walking[28]. This can be applied to the wearable lower extrem-ity exoskeleton system based on the WSM by adjustingthe stiffness of this system and, therefore, the vibrationfrequency of backpacks. Of course, when adjustingthe stiffness using this system, it should be tuned inresponse to each case by using various wearing tests ofthe vibrations of backpacks while wearers are walkingand running.

3.3.2. Selection of the spring specifications of theWSM

To implement the WSM proposed in this study, awidely known spring design method is used [29]. Whensprings are deformed, shear force, F, and torsionalmoment, T, are produced in the cross section, as shownin Fig. 5. At this time, the maximum stress loadedon the wire represents the overlapped value of shearstress by direct shear stress and torsional moment. Ifthe average diameter of springs and the diameter ofwire composing the springs are defined as D and d,respectively, this can be expressed by the strain energyanalytical method as follows:


Fig. 4. Motion strategy of the designed exoskeleton system for walking and load-carrying in the field condition.

Fig. 5. Spring design parameters.

ES = F2l

2AG+ T 2l

2GJ(2)

Here, A is the cross-sectional area and G is the shearmodulus.

Here, if the variables in Equation (2) are defined asin Equation (3), we can obtain Equation (4).

T = FD

2, l = πDNa, J = πd4

32, A = πd2

4(3)

ES = 2F2DNa

d2G+ 4F2D3Na

d4G(4)

Na in Equations (3) and (4) represents the number ofspring coils that are practically deformed in the springsand l is the length of the spring wire. If the total defor-mation of springs is defined as r, the following equationcan be derived:

r = δES

δF= 4FDNa

d2G+ 8FD3Na

d4G(5)

If we assume that the second term is greater than thefirst term, Equation (5) can be simplified as follows:

r ∼= 8FD3Na

d4G= F

ks

(6)

ks = d4G

8D3Na

(7)

As shown in the equations above, the stiffness ofsprings ks is related to the number of spring coils aswell as the external structure, such as the spring size.This finding has been used for various methods ofexoskeleton and weight-supporting systems. In otherwords, it is known that the stiffness of springs can bemodified by adjusting the number of spring coils, orby changing the shape of the springs [30, 31].


Fig. 6. Kinematic structure of the WSM.

In the remaining sections of this chapter, mathemat-ical approach will be presented for the design of theweight support mechanism for the proposed exoskele-ton system.

Figure 6 shows the mechanical assumption and mod-eling. The gravitational force induced by the mass ofthe backpack is applied at D(x, y), and the base frame islocated at point B. The position of point D is expressedas follows:

x = (l2 + l3) sin θ (8)

y = l1 − (l2 + l3) cos θ (9)

Points A and C are connected by a linear spring; itslength is expressed by

r2 = l21 + l22 − 2l1l2 cos θ (10)

The performance at point D can be expressed byusing r for each axis. Differentiating Equation (10)with time produces

rr = l1l2 sin θθ (11)

From Equations (8) and (9),

θ = x

(l2 + l3) cos θ(12)

sin θ =[

(2l1l2)2 − (l21 + l22 − r2)2

4l21l22

] 12

(13)

After Equations (12) and (13) are substituted intoEquation (11), the time-dependent displacement rela-tion between point D and spring length ls for the x-axis is

x =[

(l2 + l3)(l21 + l22 − r2)

2l21l22{

4l21l22 − (l21 + l22 − r2)2

4l21l22

}− 12

r

⎤⎦ r (14)

Briefly,

x = gxr r (15)

∵ gxr = (l2 + l3)(l21 + l22 − r2)

2l21l22{

4l21l22 − (l21 + l22 − r2)2

4l21l22

}− 12

r (16)

Likewise for the y-axis,

y =[

(l2 + l3)

l1l2r

]r (17)

y = gyr r (18)

∵ gyr = (l2 + l3)

l1l2r (19)

These equations can be used to derive the effec-tive stiffness. At point D, the effective force Feff isexpressed as follows based on the virtual work theory.

Feff = Fs + gxrFx + gy

r Fy (20)

Here, the effective stiffness of the applied spring isdefined as follows:

ks = −∂Feff

∂r= −∂Fs

∂r− gx

r

∂Fx

∂r− gy

r

∂Fy

∂r

−hxrFx − hy

r Fy

(∵ hx

r = gxr

∂r, hy

r = gyr

∂r

)(21)

For simplification, we assume that the external forceexerted on the spring mechanism is from the loadedweight on the backpack and that there is no duck-ing motion while the wearer operates this system. Ofcourse, ducking motion can be performed in the fieldoperation, and this system still support the gravitationalforce of the mass of the backpack decomposed alongthe axis parallel to the wearer’s back. However, anotherforce of the mass decomposed along the axis verticalto the wearer’s back have to be borne by wearer’s own


muscle force. It would be presented in the next chap-ter. For the standing posture, the exerted forces on eachaxis are as follows:

Fx = 0 (22)

Fy = constant (23)

If the initial length of spring is considered as r0, theelastic force is

Fs = −ks(r − r0) (24)

The effective force of Equation (20) is expressed inEquation (25) using Equations (19, 20, 22 and 24).

Feff = −ks(r − r0) + (l2 + l3)

l1l2rFy (25)

To ensure stable walking while the system is loadedwith the allowable weight, the additional condition ofFeff = 0 should be added.

ksr0 =(

ks − (l2 + l3)

l1l2Fy

)r (26)

Therefore, the equilibrium position of the springmechanism while loaded with weight is calculated asfollows:

r = ksr0

ks − (l2+l3)l1l2

Fy

(ks /= (l2 + l3)

l1l2Fy

)(27)

Selecting the stiffness of the designed spring mecha-nism involves a tradeoff between the stability of loadedweight and comfort of the wearer’s leg swing. Ade-quate stiffness of the designed system is confirmed byusing Equation (27) to maximize the allowable weight.In practice, a wearer can modify the stiffness by adjust-ing the spring mechanism while wearing the system.Finally, substituting Equation (7) into Equation (27)gives the following relationship between the springparameter of the WSM and the external load of theexoskeleton:(

d4G

8D3Na

) (l1l2

l2 + l3

) (1 − r0

r

)= Fy (28)

3.3.3. Weight balance problem while carrying thebackpack

In the case of passive exoskeleton joints of hipjoints, the moment is produced by the eccentricity withrespect to the center of the human body when sup-porting heavy personal belongings such as backpacks;

this phenomenon is often thought to adversely affectwearers’ balance while standing erect and walking.However, the following simple calculations and actualexperiments verify that this phenomenon is unlikelyto happen. The angular acceleration of the heavy per-sonal belonging of mass M in Fig. 7 can be expressedas shown in Equation (29).

θ = − g

Lsin θ (29)

Here, g is the acceleration of gravity, L is the distancefrom the center of gravity of the backpack to the centerof a passive joint of the hip joint, and θ is the rotationangle of BD with respect to AB.

The force and acceleration of mass M is expressedby the analytical method in inverted pendulum modelas follows:

Mx = Fx (30)

My = Fy − Mg (31)

x = −L(θ cos θ + θ2sinθ) (32)

y = L(θ sin θ − θ2cosθ) (33)

where θ is the angular speed of mass M with respectto passive joint B of the hip joint in Fig. 7. By sub-stituting Equation (29) into Equations (32) and (33)and multiplying each equation by the mass M of thebackpack, the magnitude of the acting force by thebackpack can be found. Finally, the component forcesexerted on the point B of the backpack in the horizontal

Fig. 7. Parameters of the WSM for the consideration of the weightbalance problem.


and gravitational directions can be derived respectivelyas follows:

−MθL cos θ − Mθ2L sin θ = Fx (34)

MθL sin θ − Mθ2L cos θ + Mg = Fy (35)

These equations are expressed in Equations (34) and(35) are simplified as follows:

M(g cos θ − Lθ2) sin θ = Fx (36)

M(g cos θ − Lθ2) cos θ = Fy (37)

If M = 40 kg, L = 0.4 m and W = 0.2 m are sub-stituted in the above equations, as the W (or θ) valuedecreases, force Fx in the direction in which the back-pack moves away from the wearer’s back drops sharply,but on the other hand force Fy in the direction ofthe acceleration of gravity increases steadily. Here,because Fy is supported by the WSM of the hip joint,as the backpack adheres more closely to the wearer’sback when its shoulder strap is tightened, Fy, which isproduced by the weight of the backpack, is minimized;this means that the wearer can maintain the balance ofthe backpack with less force. In this respect, the distri-bution of weight inside the backpack, carrying heavypersonal belongings and connected to the exoskeleton,is important [32, 33].

4. Actuation strategy and selection of theactuator specifications for knee joints of theexoskeleton system

To apply actuators to the knee joints of the exoskele-ton system, it is necessary to determine the appropriateactuator specifications by comprehensively inferringand examining the circumstances faced by wearers.This is because the increase in required torque is fol-lowed by an increase in the weight of the actuators;consequently, the increased weight of actuators actsas an additional weight on the soldiers. In existingstudies related to the selection of actuator capacity ofthe exoskeleton system, there are some cases in whichthe actuator torque capacity of the knee joints of theexoskeleton system was calculated while moving upstairs [34]. This study, in this respect, is intended tocalculate the required torque of the human body’s kneejoints while moving up and down stairs in terms of vitaldynamics, and to propose a method for transferring

power to the exoskeleton system by using the selectedactuator based on the calculation.

Existing studies indicate that the dorsi/planar flexionof ankles does not absorb a large amount of energy inthe initial stepping motion section while moving up anddown stairs and requires a large amount of force in thelatter support section [35]. Consequently, the averagedorsi/planar flexion power of ankles increases slightlywhile going up and down stairs than while walking onflat land. However, in the case of knee joints, the widestdifference in required force from that while walking onflat land occurs while moving up and down stairs; atthis time, the instantaneous maximum force capacityis known to be approximately 200 W. When comparedto 30 W when walking on flat land, this is a widevariation. The required torque in the swing section issimilar to that while walking on flat land. The requiredforce of tibiofemoral joints can be estimated by ana-lyzing photographs and films using a stroboscope formeasuring the number of rotations, and typically byanalyzing [36] the swing speed data of lower legs basedon dynamic activities such as ball kicking. Accord-ing to related studies, the maximum acceleration at theinstant of kicking a ball is 453 rad/s2, the mass momentof inertia of lower legs can be determined to be 0.35Nms2 from separate data on the human body [37], andconsequently, the torque of tibiofemoral joints can becalculated as follows:

0.35 Nms2 × 453 rad/s2 ∼= 158.5 Nm (38)

In addition to the torque, calculated by assumingthat the distance from the patellar tendon to an arbi-trary central point with respect to the tibiofemoral jointis 0.05 m, the muscular force of the knee joint, fknee,which acts through the patellar tendon, can be calcu-lated as follows (Fig. 8):

158.5 Nm = fknee × 0.05 m

∴ fkm = 3170N(39)

The above value is regarded as the force for imple-menting the maximum rotational speed of the kneejoint in the swing section. Thus, the required torqueof the knee joint of the exoskeleton for supporting themuscular power of legs while climbing stairs is thevalue obtained by subtracting the torque reaction forcein the gravitational direction, which acts on the upperlegs by the ground reaction force while stepping onthe ground, from the torque by the patellar tendon.


Fig. 8. Calculation of muscle force and external load for actuatorselection.

We assumed maximum assistive force for the groundreaction force on each foot while climbing up the stairsas 750 N based on the considered wearer’s own bodyweight. The moment of knee force is assumed thatit is always larger than the one of the ground reac-tion force regardless of the wearer’s weight and thephase of the walking cycle because there is a trade-offbetween the ground reaction force and its moment armwhile stepping up motion. Therefore, the moment ofthe ground reaction force is considered to maintain asmaller value than the moment of the knee joint for thederivation of an effective required torque τ1 for kneesupport.

τ1 = fpat × lh1 − fgr × lh2

= 3170 N × 0.05 m − 750 N × 0.15 m

∼= 50 Nm (40)

Considering the previously mentioned studies andother figures related to the instantaneous torque of thehuman body’s joints, the torque range can be deter-mined in the range of 50–113 Nm. Here, additionaltorque is required for the sheer weight of the exoskele-ton and the backpack. For a total weight (backpack,gears and one thigh segment of the exoskeleton sys-tem), M, of 65 kgf and a knee joint rotation angle,θknee, of 40◦, based on the parameters of Fig. 8 andestimated posture in phase transition from the ’weight

Fig. 9. Experimental setup of the proposed exoskeleton system.

acceptance’ to the ’pull-up’, illustrated in Fig. 12. Therequired torque of the exoskeleton knee joint, τ2, canbe calculated as follows:

τ2 = Mglexo sin θknee

= 650 N × 0.5 m × cos(70◦)

∼= 110 Nm (41)

Hence, the final torque value of the exoskeleton, bycombining Equations (40) and (41), can be calculatedas the required value in the range of approximately160 Nm. By combining the required speed value whilemoving up and down stairs (angular velocity of theknee joint), the actuator capacity (unit: W) to beapplied to the knee joint of the exoskeleton can bedetermined.

5. Manufacture of prototypes andexperimental verification

The exoskeleton system applying the designelements discussed in the previous section is imple-mented, as shown in Fig. 11; by using it, the validityof the proposed design elements is verified. The struc-ture of the exoskeleton system proposed in this studyand the result of the motion experiment of the muscle


Fig. 10. Result of the performance test of the muscle volume sensor (MVS) applied to the exoskeleton system.

volume sensor (MVS) applied to the system are shownin Figs. 9 and 10, respectively. The MVS was designedto actuate the system through the wearer’s intentionof motion; detailed structure and motion principles ofthe MVS were omitted because they digress from thesubject of this study. For further details, refer to the pre-vious research of Seo [38]. The signal patterns of theMVS when performing an isotonic motion were deter-mined by connecting a KINCOM r©, which measuresthe muscular power of each joint of the human body;the experimental subjects wearing the MVS to loadthe artificial external weight on their lower extrem-ities are shown in Fig. 10. As shown in the figure,standardized signal patterns of a certain level weredetermined with respect to various types of externalweight; these results are applicable to the exoskeletonsystem in this study, which addresses various typesof heavy personal belongings. Scenes of performingthe stair walking experiment, with the backpack of40 kgf installed in the system proposed in Fig. 9, areshown in Fig. 11. The measured muscular activitiesof the wearer before and after wearing the exoskele-ton in respective walking sections, measured by usingan electromyography (EMG) sensor attached in thevicinity of the quadriceps of the wearer’s thigh, arecompared in Fig. 12.

When wearing the exoskeleton, certain activatedmuscles can be identified in the swing phase sectionof 2–3 s; this may have been affected by the fact thatthe thigh of the exoskeleton is lifted to some degreeby the wearer’s own muscular power because the hip

joint of the exoskeleton is a hinge joint, which has noseparate power source. Of course, the state of tensionof muscles occurring during the wearer’s learning pro-cess at the initial stage of walking after wearing theexoskeleton also had some effect: the muscular activ-ity significantly dropped during the section of 5–6 s,in which the same motions are performed with respectto the same swing motion performed during the sec-tion of 2–3 s. Consequently, as the wearer passed 1 to2 cycles of walking and adapted himself to the sys-tem, which provided a certain level of muscular powerassistance for the hip joint only by actuating the motorof the knee joint through the dynamic coupling effect,which was mentioned in Subsection 2.4, could simul-taneously drop the overall muscular activity peak of thequadriceps to less than half the original level. In otherwords, it is thought that the lower extremity exoskele-ton structure in this study, which combines the WSMinstalled in the hip joint, the motor-actuated joint ofthe knee joint, and the elastic body of the ankle joint,has sufficient validity.

Muscle activation for the subjects using the devel-oped exoskeleton was fully measured and analyzed inthe previous report already. The report included fullimages of one-cycle climbing with the exoskeleton,RMS data of measured EMG signals, %MVIC of sub-jects’ muscles with/without the exoskeleton [21]. Thus,this paper additionally emphasized the result of inter-segmental, and user adaptation transition effects ofthe proposed exoskeleton while performing the weightsupporting in Fig. 12.


Stair walking experiment

Fig. 11. Evaluation of the walking performance of the exoskeleton system (40 kgf of weight loaded).

6. Conclusion and considerations

This study describes considerations for developingthe wearable lower extremity exoskeleton system forinfantry soldiers, and based on existing R&D cases,appropriate design criteria were proposed. They aresummarized in the following.

• Basically, because infantry soldiers carry a vari-ety of equipment, for an additionally worn

exoskeleton system, the interference with exist-ing equipment and additional burden on wearersshould be minimized.

• For the TJMs of hip and ankle joints and multiple-degree-of-freedom joints, the active joints of theexoskeleton are avoided, whereas the passivejoints, which are focused on the transfer and dis-tribution of the weight connected to the adjacentjoints, are applied.


Fig. 12. Measurement results of the muscular activity of the wearer’s quadriceps before/after wearing the exoskeleton system (Power assistingratio ∼= 0.5, Angular velocity of a knee joint = 30 rpm, WR: Wearable Robot).

• Walking on flat land, as shown in recent studies,can be performed by only supporting weight inthe gravitational direction; however, in the case ofmountainous terrain, active joints are necessarilyrequired for walking over rough and sloped ter-rain. For this, it is thought that the underactuatedexoskeleton system is appropriate: passive jointsare applied to hip and ankle joints and active jointsare applied to knee joints, which use the dynamiccoupling effect of the link structure and muscu-lar activation patterns while going up and downstairs.

• Although monoaxial active joints can be actuatedby applying a force sensor or muscle volume sen-sor, which has recently been introduced, becauseinfantry soldiers require various tactical motions,the responsiveness of the exoskeleton systemto the various postures and motions of wearersshould be improved by applying a foot sen-sor, clutch, and mechanism for adjusting thestiffness.

• Based on the above propositions, the prototype ofthe exoskeleton system was developed and pre-sented, and its validity was verified by a simple


stair walking experiment. The actuation perfor-mance and wearability in open fields will beverified by lightening the weight of the systemand implementing an independent power systemin further studies; the firearm rack system usingthe WSM proposed in this study will also beproposed.

References

[1] J. Knapik, E. Harman and K. Reynolds, Load carriage usingpacks: A review of physiological, biomechanical and medicalaspects, Applied ergonomics 27 (1996), 207–216.

[2] J. Bachkosky, M. Andrews, R. Douglass, J. Feigley, L. Fel-ton, F. Fernandez, P. Fratarangelo, A. Johnson-Winegar, R.Kohn, N. Polmar, R. Rumpf, J. Sommerer and W. WilliamsonLightening the Load, Arlington, VA, Naval Research AdvisoryCommittee 2007.

[3] S.A. Birrell, R.H. Hooper and R.A. Haslam, The effect of mil-itary load carriage on ground reaction forces, Gait & Posture26 (2007), 611–614.

[4] R. M. Orr, R. Pope, V. Johnston and J. Coyle, Load carriage:Minimizing soldier injuries through physical conditioning-Anarrative review, Military and Veterans’ Health 18 (2010),31–38.

[5] B.H. Jones, B.C. Hansen, K.R. Kaufman and S. Brodine,Injuries in the military: A hidden epidemic, Falls Church,VA, Armed Forces Epidemiological Board, 1996.

[6] D.N. Cowan, B.H. Jones and R.A. Shaffer, Musculoskele-tal injuries in the military training environment, Textbooks ofMilitary Medicine (2003), 195–210.

[7] http://www.lockheedmartin.com/us/products/hulc.html (Lastaccess: August 9th, 2014).

[8] http://raytheon.mediaroom.com (Last access: August 9th,2014).

[9] http://www.cyberdyne.jp (Last access: August 9th, 2014).[10] H.S. Moon, Present and Future of Wearable Robots, KISTI

report, 2012.[11] C.J. Walsh, Biomimetic Design of an Under-Actuated Leg

Exoskeleton for Load-Carrying Augmentation, PhD Thesis,Dept. of Mech. Eng, MIT, 2006.

[12] K.N. Gregorczyk, L. Hasselquist, J.M. Schiffman, C.K.Bensel, J.P. Obusek and D.J. Gutekunst, Effects of alower-body exoskeleton device on metabolic cost and gaitbiomechanics during load carriage, Ergonomics 53 (2010),1263–1275.

[13] W.S. Kim, S.H. Lee, M.S. Kang, J.S. Han and C.S. Han,Energy-efficient gait pattern generation of the poweredrobotic exoskeleton using DME, IEEE/RSJ InternationalConference on Intelligent Robots and Systems (2010),2475–2480.

[14] J.M. Donelan, R. Kram and A.D. Kuo, Mechanical work forstep-to-step transitions is a major determinant of the metaboliccost of human walking, Experimental Biology 205 (2002),3717–3727.

[15] http://www.peterpaulmuller.com/thesis/ (Last access: August9th, 2014).

[16] http://www.combatreform.org/combatlight.htm (Last access:August 9th, 2014).

[17] http://peosoldier.armylive.dodlive.mil/2010/06/28/equipment-piece-of-the-week-3/ (Last access: August 9th, 2014).

[18] http://www.army.gov.au/Our-work/Equipment-and-clothing/Soldier-Combat-Ensemble/Large-Field-Pack (Last access:August 9th, 2014)

[19] M.Q. Feng, Sensor Suits for Human Motion Detection, ReportNo: UCI-CEE-03-F03, California Univ. Irvine, 2006.

[20] S. Moromugi, Y. Koujina, S. Ariki, A. Okamoto, T. Tanaka,M.Q. Feng and T. Ishimatsu, Muscle stiffness sensor to con-trol an assistance device for the disabled, Artificial Life andRobotics 8 (2004), 42–45.

[21] S.N. Yu, H.D. Lee, S.H. Lee, W.S. Kim, J.S. Han and C.S. Han,Design of an under-actuated exoskeleton system for walk-ing assist while load carrying, Advanced Robotics 26 (2012),561–580.

[22] K.W. Hollander, P. Clouse, N. Cahill, A. Boehler, T.G. Sugarand A. Ayyar, Design of the orthotic load assistance device(OLAD), 7th Annual Dynamic Walking Conference (2012),172–173.

[23] A.E. Patla and S.D. Prentice, The role of active force andintersegmental dynamics in the control of limb trajectory overobstacles during locomotion in humans, Exp Brain Res 106(1995), 499–504.

[24] M. Bergerman, C. Lee and Y. Xu, Dynamic coupling ofunderactuated manipulators, Proceedings of the 4th IEEEConference on Control Applications, Albany, USA, (1995),500–505.

[25] Y.L. Gu and Y. Xu, Under-actuated robot systems: Dynamicinteraction and adaptive control, Systems, Man, and Cyber-netics, 1994. Humans, Information and Technology, IEEEInternational Conference on, (1994), 958–963.

[26] S.N. Yu, S.Y. Lee and C.S. Han, Methodology for thekinematical selection of a manipulator for a specified task,Autonomous Robots 23 (2007), 243–253.

[27] M.S. Cherry, Design and Evaluation of Elastic Exoskeletonsfor Human Running, PhD. Dissertation, University of Michi-gan 2010.

[28] R. Kram, Load Lugging Locomotion: Lessons from Indige-nous People, Rhino Beetles, and Wallabies, Soldier Mobility:Innovations in Load Carriage System Design and Evaluation(2001).

[29] R.C. Juvinall and K.M. Marshek, Fundamentals of machinecomponent design, Hoboken, NJ, John Wiley & Sons, 2012,Chap.12.

[30] M.H. Lee, Development of walking assistance system withadjustive leg stiffness, MS Thesis, KAIST 2011.

[31] G.W. Brown, Equipoising support apparatus, US Patent No:07618016, Issue date: Nov 17, 2009.

[32] J.P. Obusek, E.A. Harman, P.N. Frykman, C.J. Palmer andR.K. Bills, The relationship of backpack center of mass tothe metabolic cost of load carriage, Med Sci Sports Exerc 29(1997), 205.

[33] J. Knapik, Physiological, Biomechanical and MedicalAspects of Soldier Load Carriage, RTO Human Factorsand Medicine Panel (HFM) Specialists’ Meeting, Kingston,Canada, (2000), KN1-1 to KN1-14.

[34] A. Zoss and H. Kazerooni, Design of an electrically actuatedlower extremity exoskeleton, Advanced Robotics 20 (2006),967–988.

[35] J.A. Duncan, D.L. Kowalk and C.L. Vaughan, Six degree offreedom joint power in stair climbing, Gait and Posture 5(1997), 204–210.

http://www.lockheedmartin.com/us/products/hulc.html

http://raytheon.mediaroom.com

http://www.cyberdyne.jp

http://www.peterpaulmuller.com/thesis/

http://www.combatreform.org/combatlight.htm

http://peosoldier.armylive.dodlive.mil/2010/06/28/equipment-piece-of-the-week-3/

http://www.army.gov.au/Our-work/Equipment-and-clothing/Soldier-Combat-Ensemble/Large-Field-Pack

http://www.army.gov.au/Our-work/Equipment-and-clothing/Soldier-Combat-Ensemble/Large-Field-Pack


[36] V.H. Frankel and A.H. Burstein, Orthopaedic biomechanics,Lea & Febiger, 1970.

[37] R. Drillis, R. Contini and M. Bluestein, Body segment param-eters, Artificial Limbs 8 (1964), 44–66.

[38] A.R. Seo, H.Y. Jang, W.S. Kim, C.S. Han and J.S. Han, Devel-opment and verification of a volume sensor for measuringhuman behavior, Precision Engineering and manufacturing13 (2012), 899–904.

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttp://www.hindawi.com Volume 2010

RoboticsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation http://www.hindawi.com

Journal ofEngineeringVolume 2014

Submit your manuscripts athttp://www.hindawi.com

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation http://www.hindawi.com

Volume 2014

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

SensorsJournal of


Modelling & Simulation in EngineeringHindawi Publishing Corporation http://www.hindawi.com Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks


design considerations of a lower limb exoskeleton system...

Documents