Professional Documents
Culture Documents
Micromachines 2544656 Peer Review v1
Micromachines 2544656 Peer Review v1
1.School of Mechanical and Precision Instrument Engineering. Xi'an University of Technology. Xi'an 710048,
China
2. Research Center for Civil-Military Integration and Protection Equipment Design Innovation, Xi'an Univer-
sity of Technology, Xi'an 710054, China
3. State Key Laboratory of mechanical manufacturing system engineering, Xi'an Jiaotong University, Xi'an
710043, China
Abstract: In order to reduce the energy consumption of human daily movement without providing
additional power, considering the biomechanical behavior of the knee during external impedance
interaction, based on the theory of human sports biomechanics, combined with the requirements of
human-machine coupling motion consistency and coordination, an unpowered exoskeleton assisted
device for the knee joint is proposed in this paper. The effectiveness of the assisted device was ver-
ified by gait experiments and distributed plantar pressure tests with three modes, which were "not
wearing exoskeleton"(No exo.), "wearing exoskeleton with assistance "(Exo. on), and "wearing exo-
skeleton without assistance"(Exo. off). It can be clearly concluded from experimental results that: (1)
The assisted device can effectively enhance the function of the knee joint; (2) With the device, it can
reasonably improve the utilization efficiency of human energy, which is mainly reflected in collect-
ing the kinetic energy generated in the falling process and acting on the leg lifting process. (3) The
participant's lower limbs made a lot of motor adaptation to the innovative exoskeleton impedance,
which is reflected in the increase of the range of motion, force, and power of the hip and knee,
especially the decrease of knee moment, peak plantar pressure and plantar contact area. This basic
principle can be applied to assist the in-depth development of wearable devices.
Keywords: Unpowered Exoskeleton; Knee Joint; Walking Assisting; Energy Capture; Energy com-
pensation mechanism
wearer, many different strategies have been adopted. Some designs focus on weight sup-
port systems [6], while others use complex spring systems in an attempt to save energy
by reducing muscle demand [7]. Tang et al. [8] summarized that the assist components
used by the unpowered exoskeleton assist robot are usually tension spring [9-11], torsion
spring [7,12], pneumatic artificial muscle [13-15], constant force spring [16], scroll spring
[17,18], single (double) spring [19-21], spring lever and elastic cable. The existing evalua-
tion methods of exoskeleton mainly include metabolic Chen Ben, muscle activation rate
[41], joint torque, joint angle, plantar pressure and joint force. Liu et al. [23] studied the
influence of weight distribution of knee exoskeleton components on simulated muscle ac-
tivity during three functional movements.
It can be concluded from literatures above that the passive assisted exoskeleton of
the lower limb of human body is not only used as the external skeleton, but also as wear-
able equipment, which has good wearable wearability to provide protection or assistance
for the wearer. Further research and optimization are being made at home and abroad,
which can effectively solve the problem that the active device needs energy secondary
supplement, and its energy conversion efficiency also has a great space to improve. The
purpose of this study is to analyze the biomechanical characteristics and energy change
characteristics of human lower limbs. An unpowered exoskeleton assist device for knee
joint is proposed, and the gait experimental data collection and analysis are carried out.
The performance of exoskeleton is evaluated through experimental tasks in order to assist
people in walking. It can be applied to healthy people and hemiplegic patients to assist
with leg lifting, so as to reduce walking energy consumption and exercise leg walking.
Therefore, the second part gives the design principle of exoskeleton model. The third part
introduces the mathematical model of human exoskeleton system. The fourth part de-
scribes and discusses the experimental design, data processing and result analysis. In the
fifth part, we draw the conclusion of this paper.
Normal human gait usually follows a general rule, with legs repeatedly lifted and
landed in the air. When the right (left) heel begins to touch the ground and the heel
touches the ground again, this period is called a gait cycle, which is usually expressed in
seconds. As shown in Fig. 1, it shows the continuous stages after normalizing the time of
a gait cycle. The initial contact time (IC) occurs in the process of the lower limb from
swinging to standing, starting when the right (left) heel just touches the ground. In the
midstance (MS), the bodyweight is borne by the right (left) lower limbs and moves from
the heel to the forefoot. At the terminal stance (TS), the load-bearing surface of the right
(left) plantar decreases gradually, and prepare to lift the right (left) heel to step on the
ground and leave the ground. During the pre-swing (PS), the right (left) side load-bearing
foot is only in contact with the ground, and the left (right) side foot is at the moment when
the heel touches the bottom. In the initial swing (IS), the early stage is when the right (left)
side support foot leaves the ground. In the mid-swing (MS), the right (left) lower limb
retracts and approaches the left (right) lower limb, and the distance between feet reaches
the minimum value of the gait cycle. At the terminal swing (TS), the right (left) lower
limbs step forward and return to the initial gait posture.
By analyzing the structure of lower limbs, the working joints of lower limbs can be
divided into hip, knee and ankle. Each moving joint can be used as the object of energy
collection. The gravity balance state is shown in Fig. 2. It focuses on the movement of the
lower limb and simplifies the upper limb. The red rod represents the upper limb, the blue
rod represents the right lower limb, the green rod represents the left lower limb, the hol-
low circle represents the movable joint, and the solid circle represents the center of mass
of the rod. During walking, the human body is constantly doing positive and negative
work to overcome gravity. The direction of gravity 𝐺 at the shank is vertically down-
ward, and the direction of speed 𝑣 is perpendicular to the shank and points to the direc-
tion of motion. The angle 𝛼 is the angle between the direction of gravity 𝐺 and the di-
rection of velocity 𝑣. The work done by lower limb joints can be described in three cases:
(1) At 𝛼 < 90°, stretch the lower limbs and do positive work (Fig. 2 (a)); (2) At 𝛼 = 90°,
the lower limbs stand upright without doing work (Fig. 2 (b)); (3) At 𝛼 > 90°, the lower
limbs do flexion and negative work (Fig. 2 (c)).
v
v α
α
αv
G G G
(a) (b) (c)
Figure 2. Schematic diagram of gravity balance state. The lower limbs do (a) positive work, (b) no
work and (c) negative work respectively.
doing negative work. The energy capture device releases energy when the joint is doing
negative work and is in the state of doing positive work.
0%
0% 0%0%0%
0%
0% 8% 8%
8%
8%
0%8%8%8%
13% 13%
13%
13%
13%
8% 13%
30%
13% 30%30%
30%
30%
13% 30%
48%
30% 48%48%
48%
48%
30% 48%
64%
48% 64%
64%
64%
64%
48% 64%
64%74% 74%
74%
74%
74%
64% 74%
100%
74% 100%
100%
100%
100%
74% 100%
100% 100%
Knee joint power (W)
S1
0 20 40 60 80 100 X
S2
Expected curve
Actual curve
Gait cycle(%)
Figure 3. Fitting curve of the knee average power.
Y Y Y
S1<0 S1<0 S1<0
Figure 4. Power changes during assisted walking. Two conditions of joint power are described: ex-
pected curve (red) and actual curve (blue). The area formed between the two curves is indicated by
red (above the X-axis) and blue (below the X-axis.) slashes respectively. When parameter 𝑆1 is a
negative number, 𝑆2 will be (a) positive, (b) 0 and (c) negative. When parameter 𝑆1 is equal to 0,
𝑆2 will be (d) positive, (e) 0 and (f) negative. When parameter 𝑆1 is a positive number, 𝑆2 will be
(g) positive, (h) 0 and (i) negative.
It can be analyzed that the magnitude of energy absorbed and released by the energy
capture unit directly determines whether the purpose of assisted walking can be achieved,
as shown in Table 2.
pressure et al. The assisted walking device can combine the movement of the mechanical
structure with the biomechanical characteristics of human sports, so as to convert the me-
chanical energy into electrical energy or other forms of energy and feedback to the human
body. For example, the ankle exoskeleton designed and developed by Carnegie Mellon
University mentioned above, through the change of ankle angle during walking, the en-
ergy storage spring deforms to produce elastic potential energy, realizing the process of
converting mechanical energy into elastic potential energy.
It is worth mentioning that the energy capture device needs to meet two basic work-
ing states, including storing and releasing energy. In the design process of energy capture
device, the locking mechanism is added to minimize the energy consumed by the moving
joint when doing negative work, so as to ensure the minimum stress on the lower limbs
when standing vertically and obtain a high-performance energy capture device.
lt l1 lt S T
O1
θ1
mt X
m1
O2 O
ls l2
θ2
m2
φ ls
r
ms
T1
mf A
Pt
Figure 5. The human exoskeleton system. (a) State diagram of wearing lower limb exoskeleton un-
der normal walking (b) geometric relationship of lower limb parameters (c) Force analysis diagram
of scroll spring.
The zero potential energy surface S is defined by the hip joint 𝑂1 , and the gravita-
tional potential energy of the human exoskeleton system can be obtained by the following
formula
𝐸𝐺 = 𝐸𝑡ℎ𝑖𝑔ℎ + 𝐸𝑠ℎ𝑎𝑛𝑘 + 𝐸𝑓𝑒𝑒𝑡 (9)
Where, 𝐸𝑡ℎ𝑖𝑔ℎ , 𝐸𝑠ℎ𝑎𝑛𝑘 , and 𝐸𝑓𝑒𝑒𝑡 represent the gravitational potential energy of thighs,
legs and feet, respectively.
𝐸𝑡ℎ𝑖𝑔ℎ = −𝑚𝑡 𝑔𝑙𝑡′ cos 𝜃1 − 𝑚1 𝑔𝑙1 cos 𝜃1 (10)
Acoustics 2022, 4 FOR PEER REVIEW 7
Where 𝐸 is the elastic modulus of the material, 𝐼 is the moment of inertia of the material
section, and 𝑇 is the torque acting on this section. When the effective length of the scroll
spring is 𝑙, the deformation energy 𝑈 of the spring can be obtained by integrating along
the full length of the curve. The calculation is as follows
𝑙 𝑙 𝑇2 𝑇 2𝑙
𝑈 = ∫0 𝑑𝑈 = ∫0 𝑑𝑠 = (16)
2𝐸𝐼 2𝐸𝐼
After the torque 𝑇 is applied on the shaft, the scroll spring bears the same torque in
all sections of the whole length, which is equal to the torque applied on the shaft. Accord-
ing to the card theorem in Mechanics: the rate of change of the strain energy 𝑈 of the
elastic rod to a certain position 𝑃 on the rod is equal to the position 𝛿 corresponding to
the load 𝑃, which is expressed by the formula 𝛿 = 𝜕𝑈/𝜕𝑃. Therefore, the deformation
angle 𝜑 of the scroll spring can be obtained as follows.
𝜕𝑈 𝑇𝑙
𝜑= = (17)
𝜕𝑇 𝐸𝐼
The expression of torsional stiffness 𝐾 of scroll spring obtained from formulas (14)
and (17) is as follows
𝐸𝐼
𝐾= (18)
𝑙
In addition, another expression of the scroll spring energy storage formula obtained
from formulas (16) and (17) is as follows
Acoustics 2022, 4 FOR PEER REVIEW 8
1
𝑈 = 𝑇𝜑 (19)
2
If the same two scroll springs are selected in this paper, the elastic potential energy
𝐸𝑠1 = 𝐸𝑠2 = 𝑈 for spring 1 and spring 2.
The deformation angle is represented by the number of deformation turns 𝑛. Because
𝜑 = 2𝜋𝑛, the number of working turns of scroll spring is as follows
𝑇𝑙
𝑛= (20)
2𝜋𝐸𝐼
It can be calculated that 𝑑2 ≈ 0.034𝑚. The inner diameter of the spring box where
the scroll spring is placed is expressed by 𝐷2 , and the inner diameter of the coil when the
scroll spring is loose is expressed by 𝐷1 , which is obtained by the follows
Acoustics 2022, 4 FOR PEER REVIEW 9
𝐷2 ≈ 0.047𝑚 and 𝐷1 ≈ 0.034𝑚 are calculated by formulas (25) and (26). The scroll
spring is placed in the spring box, and the number of turns 𝑛1 without torque is
1 4𝑙ℎ 1
𝑛1 = (𝐷2 − √𝐷22 − )= (𝐷2 − 𝐷1 ) (27)
2ℎ 𝜋 2ℎ
The number of turns 𝑛2 of the scroll spring on the mandrel is calculated as follows
1 4𝑙ℎ 1
𝑛2 = (√ + 𝑑12 − 𝑑1 ) = (𝑑2 − 𝑑1 ) (28)
2ℎ 𝜋 2ℎ
Through the calculation of formulas (27) and (28), 𝑛1 = 7.875 cycles and 𝑛2 =
13.813 cycles are obtained.
According to the basic parameters of the scroll spring and its maximum output
torque 𝑇2 , the theoretical working revolution 𝑛𝑡 of the scroll spring can be calculated as
6𝑇2 𝑙 𝐾3 𝑙𝜎𝑏
𝑛𝑡 = = (29)
𝜋𝐸𝑏ℎ3 𝜋𝐸ℎ
Therefore, the theoretical working revolution of the spring 𝑛 ≈ 3.0601. The effective
working revolution 𝑛𝑒 of scroll spring is calculated as follows.
𝑛𝑒 = 𝐾4 (𝑛2 − 𝑛1 ) (30)
Where, 𝐾4 is the effective coefficient, and the corresponding value can be selected
through the correlation according to the ratio of 𝑑1 and 𝑛. Then 𝑛𝑒 = 5.225 turns. The
stored energy 𝑈 of the scroll spring according to formula (19).
1 1
𝑈 = 𝑇𝜑 = 𝑇2 × 2𝜋𝑛 = 𝑇2 𝜋𝑛𝑡 (31)
2 2
The energy storage density of scroll spring refers to the energy storage per unit mass,
which can be expressed by the ratio of energy 𝑈 to mass 𝑚. combined with formula (17),
the energy storage density 𝑞 of scroll spring is obtained.
𝑈 1
𝑞= = 𝑇 𝜋𝑛𝑡 (32)
𝑚 𝜌𝑉 2
Parameter Value
Thickness ℎ (𝑚𝑚) 0.8
Width 𝑏 (𝑚𝑚) 10
Torque 𝑇2 (𝑁 ∙ 𝑚) 1.689
Diameter of mandrel wound 𝑑1 (𝑚) 0.012
External diameter 𝑑2 (𝑚) 0.034
Spring coil loose ground inner diameter 𝐷1 (𝑚) 0.034
The inner diameter of spring box 𝐷2 (𝑚) 0.047
Number of turns without torque 𝑛1 (cycles) 7.875
Number of turns after spring winding 𝑛2 (cycles) 13.813
Theoretical working revolution 𝑛𝑡 (turns) 3.060
Effective working revolution 𝑛𝑒 (turns) 5.225
Acoustics 2022, 4 FOR PEER REVIEW 10
Spring2
(d)
Cable
Spring1
Bobbin
winder
Figure 6. Structure of knee joint exoskeleton. (a) The resin prototype. (b) The aluminum alloy pro-
totype. (c) Man-machine coupling of 3D digital prototype. (d) and (e) are the detail for the device.
(f) The device is on the electronic scale.
When the knee joint is extended, the displacement of the thigh and shank drives the
transmission shaft (at coil spring 1) and a rotating ring with an inelastic cable (at coil
spring 1). Then, the transmission cable rotates the rotating ring at the coil spring to drive
the other coil spring (at coil spring 1) and the ratchet pawl.
When the knee flexion in the swing phase occurs, the elastic energy stored in the knee
mounting part (at coil spring 1) is released, and the mechanical restoring force helps the
lower limb muscles lift the shank. The elastic energy in the spring of the thigh mounting
part (at coil spring 2) is also released. During a complete gait cycle, coil spring 1 helps the
muscles slow down and accelerate the knee joint, while coil spring 2 only provides similar
help during deceleration.
Through these mechanisms, when the muscle consumes metabolic energy for nega-
tive work, the knee joint exoskeleton absorbs energy. When muscles consume metabolic
energy for active work, the knee joint exoskeleton provides elastic potential energy. There-
fore, the knee exoskeleton can partially replace the function of lower limb muscles and
reduce the metabolic cost of human walking.
4. Performance evaluation
4.1. Experiment preparation and process
The proposed knee exoskeleton provides an elastic element parallel to the knee dur-
ing stance phase, but the quality and attachment mode of the exoskeleton attachment af-
fect the body state at all times. The added mass of the exoskeleton has a gravitational effect
on the hip extensor and knee flexor muscles because they must lift the exoskeleton mass
during the early swing. In addition, the increased mass also has an inertial effect on the
gluteal flexors because they must accelerate the mass during the swing. Finally, without
some restrictions, it is difficult to achieve attachment to the body, which will limit the
range of motion and cause discomfort. To assess whether exoskeletons represent an effec-
tive solution to reduce metabolic costs during weight-bearing walking. We evaluated the
performance of knee exoskeletons under three conditions, which included "not wearing
exoskeleton"(No exo.), "wearing exoskeleton with assistance "(Exo. on), and "wearing ex-
oskeleton without assistance"(Exo. off). The third case (Exo. off) is evaluated to assess the
penalties associated with the additional quality represented by the load-bearing device
itself, which is an important consideration in the design of such systems. In all trials, the
exoskeleton was worn on the right leg. Fig. 7 shows the experimental flow, data analysis
and processing flow chart.
Acoustics 2022, 4 FOR PEER REVIEW 12
In this experiment, the participant was a healthy adult male (age 23 years, height 175
cm, weight 62 kg), without joint, skeletal muscle, nerve or other diseases, and no history
of surgery. According to the experimental guidelines, inform the participant of the pur-
pose and detailed process of this experiment. After obtaining the consent of the partici-
pant, the gait acquisition experiment is carried out.
Three-dimensional (3D) motion acquisition and analysis system (VICON T40S, VI-
CON, UK), distributed plantar pressure test system (JASENCO, JSP-C5, France), and wire-
less surface electromyography test equipment (Delsys) were used in the experiment. Vi-
con optical motion capture system includes motion capture cameras (10), three-dimen-
sional force measuring platforms (3), computers, marking points, correction frames and
other auxiliary equipment, as shown in Fig. 8 (a). The shooting frequency of the motion
capture camera is 100Hz, and the data collection frequency of the force plate is 1500Hz.
Participant carried reflective marker balls and captured and recorded each marker point
on the human body by reflecting light of the same wavelength to the camera. Then
through the motion analysis software, the required motion data is collected. The distrib-
uted plantar pressure test system includes acquisition equipment, data line, computer and
data acquisition software, as shown in Fig. 8 (b).
Figure 8. Experimental equipment and marking points. (a) Motion capture experiment. (b) Distrib-
uted plantar pressure test. (c) Marker points for full body.
The improved engineering prototype was used in the experiment, as shown in Fig. 6
(b). During the experiment, the kinematics and dynamics of the joints were measured.
Participant's lower limbs used a modified Cleveland Clinic pelvis and leg marker set (left
and right ASIS, left and right greater trochanter, left and right PSI, pubic marker points,
Acoustics 2022, 4 FOR PEER REVIEW 13
thigh marker points, medial and lateral epicondyle, lower leg marker points, medial and
lateral malleolus, calcaneus, foot, fifth metatarsal) (Fig.8 (c)). After the static standing test
was calibrated, Visual 3D (CMotion Inc, Germantown, MD, USA) modeling software was
used to reconstruct the joint kinematics, dynamics and centroid trajectory, assuming that
the left and right legs were symmetrical.
Before the test, the participant was informed of the specific experimental procedure
and trained with the device for at least 30 minutes each day. Participant first trained on
the open ground, and then put on a fall-resistant seat belt on a treadmill. In this training
course, for safety reasons, the selected participant's gait should be wide enough to prevent
collision between supports, and the knee extension should ensure that the locking mech-
anism works when standing.
In the three-dimensional motion acquisition experiment (Fig.9(a)), in order to avoid
the marker ball being blocked by clothes, participant wore tight shorts and shirtless to
improve the accuracy of experimental data. Before the start of the walking experiment, it
is required to collect the static data of the participant. The participant step on a three-
dimensional force measuring board with their legs separated, and keep their arms ex-
tended. Then, the participant was asked to walk in a straight line along a prescribed route
of about 5m according to their own habits, and step on the three force plates with their
two legs one after another. During each walk, participant adjusted their pace according to
the metronome rhythm, in order to ensure that the experimental pace is uniform. Partici-
pant walked straight from the force plate and returned to the starting point of the previous
walk, counted as one walk cycle. Gait segmentation is achieved by measuring ground
reaction forces. Each walking task collected 50 walking cycles, with a rest interval of about
2 minutes between each walking cycle. Avoid large fluctuations in the collected kinemat-
ics and dynamics data due to experimental fatigue. Try to ensure the participant's natural
gait during the experiment.
RF
VL
SEM
SOL
PL
TA
Figure 9. Experimental scene. (a) Dynamic capture and (b) distributed plantar pressure test in three
states. (c) Six tested muscle positions.
During the plantar pressure dynamic test (Fig.9(b)), the participant is required to
walk through the plantar pressure measuring board with their left foot, and then walk
back with their right foot, so that the dynamic plantar pressure distribution of both feet
can be obtained. The specific experimental process is as follows. First, the participant
walked straight ahead at their chosen speed for 10 meters, with the equipment in the mid-
dle. Then, while walking on the pressure board, the participant ensured that the index
finger of their foot was on the marked white line. Finally, three successful experiments
were collected for each dynamic task, with an interval of 15 minutes between the two
experiments to eliminate the effect of fatigue.
In the process of human wearing experiment, the exoskeleton can be evaluated more
comprehensively by analyzing muscle activity. Consider the function of the muscle, the
Acoustics 2022, 4 FOR PEER REVIEW 14
depth of the muscle surface, and the position of the muscle. Delsys wireless surface elec-
tromyography test equipment was used in the experiment. The rectus femoris (RF), vastus
lateralis (VL), semitendinosus (SEM), tibialis anterior (TA), peroneus longus (PL), and so-
leus (SOL) muscle were selected for test and analysis to evaluate the assistance effect of
exoskeleton (Fig.9(c)). First, remove the hair on the skin surface with a depilator. Secondly,
the skin was wiped with 75% medical alcohol. Finally, polish the skin surface to remove
the cuticle. The direction of electrode placement is parallel to the direction of muscle fiber,
and the electrode is fixed by electrode paste. In order to prevent the EMG block from
loosening and falling off during exercise, the EMG block was wrapped with skin mem-
brane and the surface EMG signal was recorded.
Vertical axis
Frontal plane
Transverse plane
Sagittal axis
As shown in Fig. 6, the experimental flow and data analysis process are shown on
the right. The experimental data are collected through the Vicon dynamic capture test
system, and the gait data (.c3d format) file is output. Combined with Mokka software, the
preliminary data analysis of the marked points is carried out to screen out the complete
gait test data of the marked points. Importing the gait data file into Visual 3D can extract
the plantar pressure information, and select the gait motion to be analyzed during walk-
ing. The gait experimental analysis in this paper selects the gait test data from the first
landing of the right foot to the second landing of the right heel, so as to obtain the changes
Acoustics 2022, 4 FOR PEER REVIEW 15
of knee angle, force and moment. It is worth mentioning that the plantar pressure changes
on the pressure plate, then extract the plantar pressure data on the output table, and fi-
nally process the plantar pressure change curve in the gait cycle.
(a) (b)
60 No.exo. 130 Hip power 200
40 Exo.on Knee power
Exo.off 120
20 180
Knee power (W)
110
0
160
-20 100
-40 90 140
-60
80 120
-80
70
-100 100
-120 60
0 20 40 60 80 100 80
Gait cycle (%) No.exo. Exo.on Exo.off
Acoustics 2022, 4 FOR PEER REVIEW 16
Figure 11. variation of hip & knee joint power of participant in walking task. (a) The change of knee
power in three conditions during gait cycle. (b) the variation range of hip and knee power for three
conditions.
Fig.12 shows the joint angle of participant in the walking task. Fig.12(a) to Fig.12(c)
are the curves of hip angle and knee angle from the first heel contact to the second toe
contact of healthy participant in the three walking tasks. For the task of " Exo. on ", the
motion range of the knee joints was greater than that of the " Exo. off ", which increased
by about 3.72% (n=39 groups; one-way repeated-measures ANOVA; F = 14.379, P < 0.001;
LSD post hoc analysis: P < 0.001). For the task of " Exo. off ", the range of motion of the
knee joints was smaller than that of the task of "No exo.", which is reduced by about 2.05%
(LSD post hoc analysis: P = 0.005) respectively.
70 70
60 60 Exo.on
Exo.on 60 60 Exo.on
Exo.on
50 50 Exo.off
Exo.off Exo.off
Exo.off
50 50
40 40
40 40
30 30
30 30
20 20
10 10 20 20
0 0 10 10
0 010 1020 2030 3040 4050 5060 6070 7080 8090 90
100100 45 45 50 50 55 55 60 60 65 65 70 70
GaitGait
cycle
cycle
( %)( %) GaitGait
cycle
cycle
( %)( %)
Figure 12. participant's hip-knee angle in walking task. the hip-knee angle of walking task in (a)
"not wearing exoskeleton"(No exo.), (b) "wearing exoskeleton with assistance "(Exo. on), and (c)
"wearing exoskeleton without assistance"(Exo. off). (d) The change curve of knee joint angle in a
complete gait cycle. (e) Change of knee angle in 40% ~ 74% gait cycle (knee flexion to the maximum
angle).
Fig.13 shows the changes of hip &knee moment of participant in walking tasks, in-
cluding the changes of hip &knee moment from the first heel contact to the second heel
contact of healthy participant in three walking tasks (Fig.13(a)), and a box diagram of the
variation range of hip and knee moment in a complete gait cycle (Fig.13(b)). It can be
clearly seen from Fig.13(a) that the first peak value of knee moment in the "Exo. on" state
is less than that in the "Exo. off" state, which is reduced by about 8.58%, while at the second
peak value of knee joint moment, the "Exo. on" state is increased by about 25.24% com-
pared with the "Exo off" state. As can be clearly seen from Fig.13(b), for the " Exo. on" state,
the variation range of hip moment is greater than " Exo. off ", and the average joint mo-
ment increases by about 9.82% (n=39 groups; one-way repeated-measures ANOVA; F =
6.983, P = 0.003; LSD post hoc analysis: P = 0.027), while when the variation range of knee
moment is less than " Exo. off ", the average joint moment decreases by about 2.12% (n=39
groups; one-way repeated-measures ANOVA; F = 0.166, P = 0.848; LSD post hoc analysis:
P = 0.662). For the "Exo. off" state, the joint moment variation range of hip and knee is less
than that of "No exo.", and the average joint moment of hip and knee is reduced by about
13.61% (LSD post hoc analysis: P = 0.001) and 0.48% (LSD post hoc analysis: P = 0.920)
respectively. This insignificant result indicates that wearing exoskeletons does not affect
knee moment.
Acoustics 2022, 4 FOR PEER REVIEW 17
Exo.off
40
100 70
30
90 65
20
10 60
80
0 55
-10 70
50
Hip moment
-20 60 Knee moment 45
0 20 40 60 80 100
Gait cycle (%) No.exo. Exo.on Exo.off
Figure 13. Variation of joint moment of participant in the walking task. (a) The change of knee mo-
ment in three conditions during gait cycle. (b) the variation range of hip and knee moment for three
conditions.
Overall, in the dynamic gait experiment, the knee angle increased slightly (P < 0.001),
the maximum hip moment decreased (𝑃ℎ𝑖𝑝 = 0.003), the motion range of the knee joints
increased (P < 0.001), and the variation range of hip and knee power decreased (𝑃ℎ𝑖𝑝 =
0.001, 𝑃𝑘𝑛𝑒𝑒 <0.001) in the state of "Exo. on" compared with the state of "Exo. off". The ex-
perimental results show that this device has the effect of assisting walking and reducing
a certain amount of exercise energy consumption to a certain extent.
Figure 14. comparison of moving length of plantar pressure center in walking task.
Acoustics 2022, 4 FOR PEER REVIEW 18
Fig.15((a)~(d)) shows the variation diagram of plantar force, area and pressure of par-
ticipant in walking task. All data are time normalized to 101 data points. Five groups were
tested in each of the three states. The experimental data showed that for the state of "Exo.
on", the plantar force, contact area, average pressure peak and maximum pressure peak
are less than "Exo. off". For the "Exo. off" state, the plantar contact area and the maximum
pressure peak are greater than that of "No exo.", while the plantar force and average pres-
sure peak are less than that of "No exo.". Through statistical analysis (one-way ANOVA),
the experimental results are not significant (P > 0.05), which shows that the previous hy-
pothesis is not tenable. Therefore, the wearing of exoskeleton had no effect on the changes
of plantar force, area and pressure of participants. In the follow-up, we will continue to
study the effect of wearing exoskeleton on plantar parameters.
100 120
Forces - Right foot ( %)
100
80
80
Areas( cm2)
60
60
40 40
Forces-No exo. Areas - No exo.
20 Forces-Exo.on 20
Areas - Exo.on
Forces-Exo.off 0 Areas - Exo.off
0
20 40 60 80 100 20 40 60 80 100
Normalized Time ( %) Normalized Time ( %)
(c) 900 (d) 2500
800
2000
700
Max.P.( g/cm2)
Avg.P.( g/cm2)
600 1500
500
1000
400
300 Avg.P.- No exo. 500 Max.P.- No exo.
200 Avg.P.- Exo.on Max.P.- Exo.on
100 Avg.P.- Exo.off 0 Max.P.- Exo.off
20 40 60 80 100 20 40 60 80 100
Normalized Time ( %) Normalized Time ( %)
Figure 15. Variation of plantar force, areas and pressure in walking task. (a), (b), (c), and (d) describe
the changes in plantar forces, contact areas, average pressure, and maximum pressure, respectively.
The time is normalized.
It can be seen that in the dynamic distributed plantar pressure experiment, the COP
length in the "Exo.on" state is reduced, which compared with "Exo.off" state (P = 0.027).
However, compared with the state of "No exo.", the state of "Exo. off" shows that the plan-
tar contact area and the maximum pressure peak increase. On the one hand, by reducing
the weight of the exoskeleton, we can reduce the load of the human lower limb wearing
the exoskeleton and continuously improve the assistance performance of the exoskeleton
to the lower limb. On the other hand, further increase the static test experiment, and fur-
ther verify that the assisted exoskeleton can reduce the energy consumption of lower
limbs from the physiological level through lower limb electromyography test and metab-
olism test.
TA, and PL are more than that of "No exo.". The EMG test results showed that when par-
ticipant wore exoskeleton to assist walking, most EMG signals decreased slightly except
semitendinosus, tibialis anterior, and peroneus longus muscle. This is because the exo-
skeleton is the external structure of the body, which causes additional muscle metabolism
to do work and consumes energy when walking.
We met with mixed success in our objectives. These results suggest that the tibialis
anterior muscle can be assisted during the initial gait swing, although the semitendinosus
muscle has intense activity during the swing. Part of the reason may be caused by wearing
problems during the experiment. In addition, there may be some differences in results
due to differences in methods.
In conclusion, through the analysis of the experimental results, it can be used to eval-
uate the impact of exoskeleton on human gait, plantar pressure and muscle activation.
Compare the evaluation indexes of the proposed exoskeleton with the existing unpow-
ered exoskeleton of the lower limbs, as shown in Table 4. For example, Banala, S.K. et al.
[28] proposed a multi joint (hip and knee) gravity balanced leg orthosis. The angles of hip
and knee joints of healthy wearers increased, and the EMG activities of rectus femoris and
hamstring muscles decreased. Similarly, the motion range of hip and knee angle in this
paper increases, RF activity decreases, but SEM activity increases. Generally, in order to
improve the assisted efficiency of exoskeleton, the power will be added to the unpowered
exoskeleton for the operable electronic control system, clutch or variable shock absorber
[29]. This will be our next research direction of wearable assisted exoskeleton.
(a) 4.5 No exo. Exo.on (b) 6.0 No exo. Exo.on (c) 1.8 No exo. Exo.on
4.0 5.5 1.6
3.5 5.0 1.4
SEM (10-5V)
RF (10-5V)
VL (10-5V)
8
PL (10-5V)
25 3.0
TA (10-5V)
2.5 7
20 6
2.0
15 5
1.5
10 4
1.0
5 3
0.5 2
0 0.0 0 100
0 20 40 60 80 100 20 40 60 80 0 20 40 60 80 100
Gait cycle (%) Gait cycle (%) Gait cycle (%)
Figure 16. Muscle activation. These curves represent two different conditions: “No exo.” (red) and
“Exo. on” (cyan). During the gait cycle (heel to heel contact), the active state of six tested muscles
was analyzed, including (a) the rectus femoris, (b) vastus lateralis, (c) semitendinosus, (d) tibialis
anterior, (e) peroneus longus, and (f) soleus muscle.
Ref [18,32] K. -- -- -- -- -- -- -- -- -- -- ↓
Ref [33] K. -- -- -- -- -- -- -- -- ↓ -- --
Ref [17] K. -- -- -- -- -- -- ↓ -- -- -- --
This work K. ↑ ↑ ↑ ↑ ↑ -- ↓ ↑ -- ↓ --
↑ represents an increase after wearing; ↓ represents a decrease after wearing; 𝑃ℎ : Hip power; 𝑃𝑘 :
Knee power; 𝐴ℎ : Hip angle; 𝐴𝑘 : Knee angle; 𝑀ℎ : Hip moment; 𝑀𝑘 : Knee moment; PF: Plantar force;
GRF: Ground reaction forces; L.COP: Length of COP (center of pressure); Met.: Metabolic; H.K.: Hip
and knee; H.K.A.: Hip, knee and ankle; H.: Hip; K.: Knee.
6. Conclusion
In this paper, an unpowered knee exoskeleton is proposed, which is used to reduce
the energy consumption of lower limbs based on the energy compensation mechanism,
so as to achieve the purpose of assisted walking. The test prototype and engineering pro-
totype are made with resin and aluminum alloy. The gait experiments and plantar pres-
sure tests are carried out with an aluminum prototype to verify its effectiveness. In the
experiment, the kinematic and dynamic data of hip & knee, as well as plantar pressure
characteristics, were collected and analyzed. The experimental results show that: (1) For
the task of " Exo. on ", the variation range of hip and knee power, knee angle, and hip
moment was greater than that of the " Exo. off ", which increased by about 21.13% (P =
0.001), 39.07% (P < 0.001), 3.72% (P < 0.001), and 9.82% (P=0.003) respectively. But the mov-
ing length of plantar pressure center is less than that of "Exo. off", which is reduced by
about 6.57% (P = 0.027). (2) For the "Exo. off" state, the range of hip and knee power, knee
angles, and hip moment are less than the "No exo." state, which reduced by 21.60% (P =
0.001) and 21.69% (P < 0.001), 2.05% (P < 0.001), 13.61% (P = 0.003) respectively. But the
moving length of plantar pressure center is more than that of "Exo. off", which is increased
by about 5.38% (P = 0.027). (3) the tibialis anterior muscle can be assisted during the initial
gait swing, although the semitendinosus muscle has intense activity during the swing. It
should be emphasized that the energy consumption and muscle activation of lower limbs
need to be further verified by a large number of subsequent experiments. It is believed
that this assist device can become a potential orthosis for stroke patients and can be used
as an auxiliary rehabilitation device for patients with muscle weakness.
Acknowledgments:This work was supported by Ministry of education Youth Fund Project (Grant
numbers 21XJC760003), Outstanding Talents Support Program Project (Grant numbers 106-
451420001), and Common technology and field fund for equipment pre research (Grant numbers
106-418321001). The authors would like to express their appreciation to the agencies.
References
1. Westerterp KR. Physical activity and physical activity induced energy expenditure in humans: measurement, determinants,
and effects. Front Physiol. 2013; 4: 90.
2. Pont-Esteban D, Sánchez-Urán M Á , Ferre M. Robust Motion control architecture for an upper-limb rehabilitation Exosuit, IEEE
Access, 2022, 10(10): 113631-113648.
3. Zhang Y, Bressel M, De Groof S, et al. Design and control of a size-adjustable pediatric lower-limb exoskeleton based on weight
shift, IEEE Access, 2023, 11(11): 6372-6384.
4. Kwa H K, Noorden J H, Missel M, et al. Development of the IHMC mobility assist exoskeleton[C]//2009 IEEE international
conference on robotics and automation. IEEE, 2009: 2556-2562.
5. Zoss A B, Kazerooni H, Chu A. Biomechanical design of the Berkeley lower extremity exoskeleton (BLEEX)[J]. IEEE/ASME
Transactions on mechatronics, 2006, 11(2): 128-138.
6. Lee K M, Wang D. Design analysis of a passive weight-support lower-extremity-exoskeleton with compliant knee-joint[C]//2015
IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2015: 5572-5577.
7. Van Dijk W, Van der Kooij H. XPED2: A passive exoskeleton with artificial tendons[J]. IEEE robotics & automation magazine,
2014, 21(4): 56-61.
8. Tang, X.; Wang, X.; Ji, X.; Zhou, Y.; Yang, J.; Wei, Y.; Zhang, W. A Wearable Lower Limb Exoskeleton: Reducing the Energy
Cost of Human Movement. Micromachines 2022, 13, 900.
9. Agrawal S K, Banala S K, Fattah A, et al. Assessment of motion of a swing leg and gait rehabilitation with a gravity balancing
exoskeleton[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2007, 15(3): 410-420.
Acoustics 2022, 4 FOR PEER REVIEW 21
10. Zhou L, Chen W, Chen W, et al. Design of a passive lower limb exoskeleton for walking assistance with gravity compensation[J].
Mechanism and Machine Theory, 2020, 150: 103840.
11. Guan X, Ji L, Wang R, et al. Optimization of an unpowered energy-stored exoskeleton for patients with spinal cord in-
jury[C]//2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE,
2016: 5030-5033.
12. Chaichaowarat R, Kinugawa J, Kosuge K. Unpowered knee exoskeleton reduces quadriceps activity during cycling[J]. Engi-
neering, 2018, 4(4): 471-478.
13. Yandell M B, Tacca J R, Zelik K E. Design of a low profile, unpowered ankle exoskeleton that fits under clothes: overcoming
practical barriers to widespread societal adoption[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering,
2019, 27(4): 712-723.
14. Leclair J.Development and testing of an unpowered ankle exoskeleton for walking assist [D]. Université d'Ottawa/University
of Ottawa, 2016: 1-125.
15. Leclair J, Pardoel S, Helal A, et al. Development of an unpowered ankle exoskeleton for walking assist[J]. Disability and Reha-
bilitation: Assistive Technology, 2018.
16. Shepertycky M, Burton S, Dickson A, et al. Removing energy with an exoskeleton reduces the metabolic cost of walking[J].
Science, 2021, 372(6545): 957-960.
17. Ranaweera R, Gopura R, Jayawardena T S S, et al. Development of A Passively Powered Knee Exoskeleton for Squat Lifting[J].
J. Robotics Netw. Artif. Life, 2018, 5(1): 45-51.
18. Xie L, Huang G, Huang L, et al. An unpowered flexible lower limb exoskeleton: Walking assisting and energy harvesting[J].
IEEE/ASME Transactions on Mechatronics, 2019, 24(5): 2236-2247.
19. Grabowski A M, Herr H M. Leg exoskeleton reduces the metabolic cost of human hopping[J]. Journal of Applied Physiology,
2009, 107(3): 670-678.
20. Van Wijdeven M A P. Design and evaluation of a passive hip exoskeleton to reduce the energy cost of human walking[D]. Delft
: Delft University of Technology, 2016: 13-50.
21. Nasiri R, Ahmadi A, Ahmadabadi M N. Reducing the energy cost of human running using an unpowered exoskeleton[J]. IEEE
Transactions on Neural Systems and Rehabilitation Engineering, 2018, 26(10): 2026-2032.
22. Zhang, P.; Wu, P.; Wang, W. Research on Lower Limb Step Speed Recognition Method Based on Electromyography. Microm-
achines 2023, 14, 546.
23. Liu Y X, Zhang L, Wang R, et al. Weight distribution of a knee exoskeleton influences muscle activities during movements, IEEE
Access, 2021, 9(9): 91614-91624.
24. Bovi G, Rabuffetti M, Mazzoleni P, et al. A multiple-task gait analysis approach: kinematic, kinetic and EMG reference data for
healthy young and adult subjects[J]. Gait & posture, 2011, 33(1): 6-13.
25. Duan W, Feng H, Wang Z Q. Finite element analysis of flat spiral spring in elastic energy storage device [J]. Chinese Journal of
Construction Machinery, 2011, 9(4): 493-498.
26. Jiang Y, Zhang Y H, Wan G X, et al. Design calculation of flat spiral spring [M]. Beijing: Machinery Industry Press, 1994.
27. Silva M P T, Ambrósio J A C. Sensitivity of the results produced by the inverse dynamic analysis of a human stride to perturbed
input data[J]. Gait & posture, 2004, 19(1): 35-49.
28. Banala, S.K.; Agrawal, S.K.; Fattah, A.; Krishnamoorthy, V.; Hsu, W.-L.; Scholz, J.; Rudolph, K. Gravity-Balancing Leg Orthosis
and Its Performance Evaluation. IEEE Trans. Robot. 2006, 22, 1228–1239.
29. Ma, J.; Sun, D.; Ding, Y.; Luo, D.; Chen, X. Cooperativity Model for Improving the Walking-Assistance Efficiency of the Exo-
skeleton. Micromachines 2022, 13, 1154.
30. Van der Wijdeven, M.A.P. Design and Evaluation of a Passive Hip Exoskeleton to Reduce the Energy Cost of Human Walking;
Delft University of Technology, Dept. of Mechanical Engineering: Delft, The Netherlands, 2016.
31. Yuan, B.; Li, B.; Chen, Y.; Tan, B.; Jiang, M.; Tang, S.; Wei, Y.; Wang, Z.; Ma, B.; Huang, J. Designing of a Passive Knee-Assisting
Exoskeleton for Weight-Bearing. In Proceedings of theInternational Conference on Intelligent Robotics and Applications, Wu-
han, China, 6 August 2017; pp. 273–285.
32. Shen, J.; Zhang, S.; Chen, C.; Xu, Z. Research Progress of Unpowered Exoskeleton Assist Robot. J. Mech. Transm. 2020, 44, 166–
176.
33. Kim, H.; Lim, D.; Kim, W.; Han, C. Development of a Passive Modular Knee Mechanism for a Lower Limb Exoskeleton Robot
and Its Effectiveness in the Workplace. Int. J. Precis. Eng. Manuf. 2019, 21, 227–236.