Development of Concentric-only Exercise Machine with Estimation of
Whole Body Dynamics
Toyoyuki Honjo
1
, Naruhiro Shiozawa
2
, Seiichi Yokoi
2
and Tadao Isaka
2
1
Ritsumeikan Global Innovation Research Organization, Ritsumeikan University,
1-1-1, Nojihigashi, Kusatsu, Shiga, 525-8577, Japan
2
Graduate School of Sport and Health Science, Ritsumeikan University,
1-1-1, Nojihigashi, Kusatsu, Shiga, 525-8577, Japan
Keywords:
Training Machine, Lower Extremity Exercise, Concentric Exercise, Whole Body Dynamics.
Abstract:
Recently, the effects of concentric and eccentric training have been evaluated to determine how to make exer-
cise more effective. Some researchers have reported that concentric exercise increases the concentric strength.
In this case, concentric exercise may enhance the concentric force performance with or without lower muscle
damage and pain. Therefore, we developed a concentric training machine for lower extremities that we call
iSAAC. This machine has an electromagnetic brake to generate a safe resistance load for exercise. The mag-
nitude of the resistance power is less than or equal to the human-applied power. We calculated whole body
dynamics such as the joint torque and work of the lower extremities based on the inverse dynamics of the
whole body without constraints. We verified the effectiveness of the system through squat-like knee extension
exercise and inverse dynamics analysis. The proposed machine provided safe training for concentric knee
extension and information on the whole body dynamics during exercise.
1 INTRODUCTION
Daily exercise is important for people to keep or im-
prove their motor ability. Many people go to fitness
gyms and use the latest training machines. The merit
of machine training is that these machines let the user
utilize a heavy training load. Various intelligent ex-
ercise machines have been developed to enhance the
training effect on specific motor abilities.
Concentric and eccentric power generation are im-
portant factors in deciding a training menu. Some
researchers have reported that eccentric training in-
creases muscle hypertrophy or muscle damage (Seger
et al., 1998; Gonzalez-Izal et al., 2014). Others have
reported that there is no significant difference in re-
sults between concentric and eccentric training (Ben-
Sira et al., 1995). Blazevich et al. and Cadore et
al. reported that a concentric training group showed
a greater increase in the concentric torque than an ec-
centric training group (Blazevich et al., 2007; Cadore
et al., 2014). Cadore et al. reported that the eccentric
training group showed a greater increase in the eccen-
tric torque than the concentric exercise group (Cadore
et al., 2014), whereas Blazevich et al. reported that
there was no significant difference in the increase in
eccentric torque between the concentric and eccentric
training groups (Blazevich et al., 2007). These results
show that concentric exercise may enhance the con-
centric force performance with or without lower mus-
cle damage and pain. Therefore, the effect of concen-
tric training alone needs to be verified.
In general, most traditional cyclic resistance train-
ing includes both concentric and eccentric power gen-
eration. Therefore, a special exercise machine is re-
quired for cyclic unilateral concentric or eccentric re-
sistance training.
In this paper, we introduce a concentric exercise
machine test bed based on a passive power supply.
We call this the Intelligent System of Advanced Ac-
tuation for Concentric training (iSAAC). The iSAAC
adopts electromagnetic brakes as a power source for
the training load. Many researchers have focused on
safety in the development of machines that will inter-
act with humans because the machine must not harm
the human body. We selected a brake as the safety
actuation device. The magnitude of the reaction force
is less than or equal to the applied force, and its di-
rection is opposite to that of the motion. Therefore,
the braking force is suited for concentric training re-
sistance.
170
Honjo T., Shiozawa N., Yokoi S. and Isaka T..
Development of Concentric-only Exercise Machine with Estimation of Whole Body Dynamics.
DOI: 10.5220/0005142101700176
In Proceedings of the 2nd International Congress on Sports Sciences Research and Technology Support (icSPORTS-2014), pages 170-176
ISBN: 978-989-758-057-4
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
(a) iSAAC.
(b) Base unit.
(c) L-shaped unit. (d) Sliding board.
Figure 1: Image of iSAAC and its principal composition elements.
Second, a training machine can provide the user
with a unique resistance load. Many developed ma-
chines target the exercise of a single joint, such as the
knee or elbow, for safety or portability training (Shu-
fang et al., 2006; Kikuchi et al., 2009; Nikitczuk et al.,
2010). Moromugi et al. proposed a multi-joint exer-
cise machine based on a muscle stiffness sensor (Mo-
romugi et al., 2006). The resistance load is generated
via sensory information from the specific muscle ac-
tivity. The main target of our system is multi-joint
exercises for knee extension, such as squat exercises.
Therefore, the training load is determined by the knee
angle during exercise.
This machine has a movable seat, where the user
lies on his or her back. The user’s trunk is fixed to
the seat by two belts. The electromagnetic brake ap-
plies a resistance force to the trunk through this seat.
Most sensors are embedded in the machine; the user
only wears a goniometer. The lower body is not con-
strained during exercise. Therefore, the user can read-
ily and comfortably use this machine.
We verified the effectiveness of this system
through squat exercises and analysis of the whole
body dynamics. The results showed that this system
provides cyclic concentric knee extension exercise.
2 DETAILS OF iSAAC
2.1 Structure
Figure 1(a) shows an image of the iSAAC. The
iSAAC targets the muscles for knee extension based
only on concentric power generation. This machine
consists of a base unit, an L-shaped slope unit, a
sliding board with wheels, various sensors, and an
electromagnetic brake. The base unit has two pil-
lars. These pillars support the slope unit at 30
◦
, 45
◦
,
or 60
◦
. The L-shaped slope unit has a force plate
(Kistler, 9281CA), an electromagnetic brake (Mit-
subishi, ZA20Y1), two pulleys, a timing belt, and a
rotary encoder. The under pulley is connected through
a shaft to the electromagnetic brake and is connected
DevelopmentofConcentric-onlyExerciseMachinewithEstimationofWholeBodyDynamics
171
Figure 2: Image of lower extremity training.
through a timing belt to the upper pulley. The force
plate is fixed to the bottom of the slope unit. The slope
unit has a pair of rails consisting of two parallel pipes.
The sliding board has six wheels, two seat belts, and
one connectivedevice. The sliding board is located on
these rails and is connected with the timing belt. The
user lies on his or her back on the seat, and the user’s
trunk is fixed on the sliding board by two belts, as
shown in Figure. 2. The braking torque from the elec-
tromagnetic brake acts as the resistance power when
the sliding board slides on the rails.
2.2 Actuators
The iSAAC adopts the braking force as training resis-
tance. The merit of using the electromagnetic brake as
training resistance is the safety for the human body.
The braking resistance acts as a training load while
the user lifts up his or her body. The resistance force
from the electromagnetic brake is less than or equal to
the human-generated force. Therefore, the following
relations are satisfied:
|F
EB
| =
|F
H
| (|F
H
| < |F
TL
|)
|F
TL
| (|F
H
| ≥ |F
TL
|)
(1)
where F
EB
is the resistance force, F
H
is the human-
generated force, and F
TL
is the setup value from the
control PC. This is a great advantage because this sys-
tem automatically regulates the resistance without any
delay.
As shown in Figure. 3, the electromagnetic brake
is controlled by the input from a DC power supply
(Takasago Electric Industry, KX-100L). KX-100L is
connected to a control PC by an RS-232C cable with
the serial communication method. The control or-
ders for KX-100L are created by LabVIEW (National
Instruments, LabVIEW 8.2). We can only control
the braking torque generated by the electromagnetic
brake. Therefore, this machine has no active power
Figure 3: System configuration diagram.
Figure 4: Data flow diagram for dynamics analysis.
Two-pass fourth-order Butterworth filter was adopted for
smoothing. The cutoff frequency was 12 Hz. The sampling
rate of the inverse dynamics was 1 kHz.
supply device. The rated torque of this electromag-
netic brake is 200 Nm. The radius of the pulleys is
0.0980 m. Therefore, the maximum training resis-
tance is about 208 kg.
2.3 Sensors
The iSAAC has three sensors. The first sensor is a
force plate (Kistler Co., Ltd., 9281CA) located on the
bottom of the L-shaped unit and measures the ground
reaction force during exercise. The second sensor is a
rotary encoder (Koyo Electronics Industries Co., Ltd.,
TRD-2E1000B) mounted on the shaft end of the up-
per pulley to measure the locomotion of the sliding
board. The last sensor is a goniometer (Biometrics
Ltd., SG150) that measures three joint angles (hip,
knee, and ankle joints) on the sagittal plane. The an-
gles are relative angles from the proximal segment to
the distal segment. The trunk angle is equal to the
machine slope. Sensory signals such as the ground
reaction forces, pulley rotation, and joint angles are
sent to the control PC through A/D boards (DKH,
PH701, and PH710) and the PCI board (National In-
struments, PCI-6225). All sensor signals are acquired
and recorded by the LabVIEW program through the
PCI board. The sampling rate of data acquisition is 4
kHz.
icSPORTS2014-InternationalCongressonSportSciencesResearchandTechnologySupport
172
(a) Configuration of physical parameters for segment. (b) Joint angle configuration.
Figure 5: Four rigid bodies of human model for inverse dynamics analysis. The first segment is the trunk, which includes the
head, both arms, and the pelvis. The trunk is fixed to the sliding board. The second segment is the thigh, which is connected to
the trunk by the hip joint. The third segment is the shank, which is connected to the thigh by the knee joint. The last segment
is the foot, which is connected to the shank by the ankle joint and is in contact with the force plate. l
i
, (i = 0, ·· · , 3) are the
segment lengths. a
i
, (i = 0, · ·· , 3) are the distances from the joint to each segment’s center of mass. m
i
, (i = 0, · ·· , 3) are the
segment masses, and I
i
, (i = 1, ··· , 3) are the segments’ inertia moments. The inertia moment of the trunk has no effect on the
motion equation (2) because the trunk does not rotate.
(a) Linear behavior of trunk along slope. (b) Joint angles of lower extremity.
Figure 6: Measured kinetic data during exercise.
3 ANALYSIS OF DYNAMICS
We implemented the analysis of dynamics that cal-
culates the joint torque of the lower extremity on the
sagittal plane. The analysis system requires sensory
information on the motion data and external forces
from the environment, as shown in Figure. 4. We
assumed that the behavior of the left leg is synchro-
nized with the behavior of the right leg to reduce the
number of sensors. Therefore, the human model is
represented by four rigid segments, as shown in Fig-
ure. 5.
The motion equation of this model is described by
M(q)¨q+ h(q, ˙q) = S
T
τ+ J
T
R
F
R
+ J
T
GRF
(q)F
GRF
. (2)
where q = [p+ p
of fset
, θ
1
, θ
2
, θ
3
]
T
∈ R
4
is the gen-
eralized coordinate vector; M ∈ R
4×4
is the inertia
matrix; and h ∈ R
4
is the vector that includes the
Coriolis force, centrifugal force, and gravity term.
τ = [τ
1
, τ
2
, τ
3
]
T
∈ R
3
is the joint torque vector, and
S
T
∈ R
4×3
is its Jacobian matrix. J
T
R
F
R
∈ R
4
is the
resistance force. F
R
is the resistance force in the di-
rection of the slope and includes the effect of grav-
ity on the sliding board. J
T
GRF
(q)F
GRF
∈ R
4
is the
ground reaction force. F
GRF
= [F
GRFx
, F
GRFz
]
T
∈ R
2
is the total ground reaction forces of both the right and
left feet. p
of fset
is the constant offset position of the
sternum when the sliding board is located at the low-
est (initial) point. p is the displacement of the sliding
board and the trunk from p
of fset
. θ
1
, θ
2
, θ
3
are the
angles of the hip, knee, and ankle joints, respectively.
As shown in Figure. 4, p, θ
1
, θ
2
, θ
3
are calculated
based on sensory information. ˙q = [ ˙p,
˙
θ
1
,
˙
θ
2
,
˙
θ
3
]
T
and
¨q = [ ¨p,
¨
θ
1
,
¨
θ
2
,
¨
θ
3
]
T
are numerically derived after the
filtering process.
Physical parameters such as the inertia moment of
the segment are derived from the body weight and
segment length from the previous study (Ae et al.,
1992); these values are shown in Table 1. The term
of the joint torque and the term of the resistance force
DevelopmentofConcentric-onlyExerciseMachinewithEstimationofWholeBodyDynamics
173
(a) Joint torque of lower extremity. (b) Joint work of lower extremity.
(c) Ground reaction force. (d) Resistance force.
Figure 7: Analyzed dynamic data of squat-like exercise.
Table 1: Physical parameters of human model.
Name Value Unit
M
g
67.000 [kg]
M
SB
27.000 [kg]
m
0
0.656M
g
[kg]
m
1
0.220M
g
[kg]
m
2
0.102M
g
[kg]
m
3
0.022M
g
[kg]
l
0
0.500 [m]
l
1
0.390 [m]
l
2
0.390 [m]
l
3
0.260 [m]
a
0
0.493l
0
[m]
a
1
0.475l
1
[m]
a
2
0.405l
2
[m]
a
3
0.406l
3
[m]
I
1
m
1
(0.278l
1
)
2
[kgm
2
]
I
2
m
2
(0.274l
2
)
2
[kgm
2
]
I
3
m
3
(0.177l
3
)
2
[kgm
2
]
are normal to each other. Therefore, the joint torque
and resistance force are derived as follows:
τ =
SS
T
−1
S
M(q) ¨q+ h(q, ˙q) − J
T
GRF
(q)F
GRF
(3)
and
F
R
=
J
R
J
T
R
−1
J
R
[M(q) ¨q+ h(q, ˙q)
−S
T
τ− J
T
GRF
(q)F
GRF
.
(4)
4 EXPERIMENTS AND
DISCUSSION
In this study, we asked a subjects to perform a cyclic
concentricsquat-like exercise using the iSAAC to ver-
ify the effectiveness of this system. The subject per-
formed squat-like exercises with a constant braking
torque. The slope angle θ
0
was 30
◦
. His resistance
force of the electromagnetic brake was 115 Nm =
1173 N. This was equal to 70% of one rep max (1RM)
for the subject.
The controller of the squat-like exercise was im-
plemented as follows. First, the electromagnetic
brake exerted a resistance force when the subject ex-
tended his knee (θ
2
≤ −10
◦
). Second, the electro-
magnetic brake did not operate when the subject ex-
tended his knee (− 10
◦
≤ θ
2
) or bent his knee after
knee extension. Finally, the electromagnetic brake
again exerted the resistance force when the subject
deeply bent his knee (θ
2
≤ −80
◦
) after knee exten-
icSPORTS2014-InternationalCongressonSportSciencesResearchandTechnologySupport
174
Figure 8: Animation of exercise from sitting position to standing position.
sion. The controller repeated the above three steps.
Figures 6 and 7 show the measured and analyzed
data of the squat exercises for the first 30 s. A cyclic
concentric exercise was achieved with the iSAAC,
as shown in Figure. 6. The resistance force acted
as training resistance when the subject extended his
knee. In addition, the braking force supported the
trunk when the subject sat. The joints exerted the
extension torque during the exercise and did positive
work, as shown in Figures. 7(a) and (b). Positive joint
work means the concentric power generation. There-
fore, these results showed that this system lets the user
perform safe cyclic concentric training of the knee
extension. The resistance force mostly matched the
summation of 70% 1RM (1173 N) and the effect of
gravity on the sliding board (M
SB
gsin|θ
0
| = 132.3N),
as shown in Figure. 7(d). The system provided the
suitable resistance load. Resultant GRF was similar
to that of squat exercise (Kellis et al., 2005). Peak
knee and ankle joint torque were similar to those of
motor controlled leg press (Hahn et al., 2010). Peak
hip and ankle joint torque were smaller than those of
full squat exercise (80% of 1RM) (Robertson et al.,
2008). However, peak knee torque was larger than
that of full squat exercise (80% of 1RM) (Robert-
son et al., 2008). Therefore, the system provided the
squat-like knee extension training. In addition, these
results show that iSAAC could estimate adequate dy-
namics. Figure 8 shows the animation of one knee
extension movement.
5 CONCLUSIONS
In this study, we developed a concentric training ma-
chine where the training load is generated by a brak-
ing torque. The braking torque acts as concentric
training resistance while the user lifts up his or her
body. We verified the effectiveness of this machine
through squatting exercises and analysis of the inverse
dynamics. The results showed that this system pro-
vides concentric leg extension exercise. In the future,
we will investigate the effects of short- and long-term
training programs based on the iSAAC.
ACKNOWLEDGEMENTS
This work was supported by a grant from JSPS KAK-
ENHI (No. 24300220).
DevelopmentofConcentric-onlyExerciseMachinewithEstimationofWholeBodyDynamics
175
REFERENCES
Ae, M., Tang, H., and Yokoi, T. (1992). Estimation of in-
ertia properties of the body segments in japanese ath-
letes. Society of Biomechanisms Japan, 11:23–33. (in
Japanese).
Ben-Sira, D., Ayalon, A., and Tavi, M. (1995). The effect
of different types of strength training on concentric
strength in women. Journal of Strength and Condi-
tioning Research, 9(3):143–148.
Blazevich, A., Cannavan, D., Coleman, D., and Horne,
S. (2007). Influence of concentric and eccentric re-
sistance training on architectural adaptation in hu-
man quadriceps muscles. Journal of Applied Physi-
ology, 103(5):1565–1575. DOI: 10.1152/japplphys-
iol.00578.
Cadore, E., Gonzalez-Izal, M., Pallares, J., Rodriguez-
Falces, J., Hakkinen, K., Kraemer, W., Pinto, R., and
Izquierdo, M. (2014). Muscle conduction velocity,
strength, neural activity, and morphological changes
after eccentric and concentric training. Scandina-
vian Journal of Medicine and Science In Sports. doi:
10.1111/sms.12186.
Gonzalez-Izal, M., Cadore, E., and Izquierdo, M. (2014).
Muscle condition velocity, surface electromyography
variables, and echo intensity during concentric and ec-
centric fatigue. Muscle and Nerve, 49(3):389–397.
Hahn, D., Seiberl, W., Schmidt, S., Schweizer, K., and
Schwirtz, A. (2010). Evidence of residual force en-
hancement for multi-joint leg extension. Journal of
Biomechanics, 43(8):1503–1508.
Kellis, E., Arambatzi, F., and Papadopoulos, C. (2005). Ef-
fects of load on ground reaction force and lower limb
kinematics during concentric squats. Journal of Sports
Sciences, 23(10):1045–1055.
Kikuchi, T., Oda, K., Ohyama, Y., Isozumi, S., and Furusho,
J. (2009). Development of isokinetic exercise system
using high performance mr fluid brake. IEEE Interna-
tional Conference on Mechatronics.
Moromugi, S., Yoon, S., Kim, S., Tanaka, M., Ohgiya, Y.,
Matsuzaka, N., and Ishimatsu, T. (2006). A training
machine with dynamic load-control function based
on muscle activity information. Artificial Life and
Robotics, 10(2):126–130.
Nikitczuk, J., Weinberg, B., Canavan, P., and Mavroidis, C.
(2010). Active knee rehabilitation orthotic device with
variable damping characteristics implemented via an
electrorheological fluid. IEEE/ASME Transactions on
Mechatronics, 15(6):952–960.
Robertson, D., Wilsom, J., and Pierre, T. (2008). Lower
extremity muscle functions during full squats. Journal
of Applied Biomechanics, 24(4):333–339.
Seger, J., Arvidsson, B., and Thorstensson, A. (1998). Spe-
cific effects of eccentric and concentric trainig on
muscle strength and morphology in humans. Euro-
pean Journal of Applied Physiology, 79:49–57.
Shufang, D., Lu, K., Sun, J., and Rudolph, K. (2006). Adap-
tive force regulation of muscle strengthening rehabil-
itation device with magnetorheological fluids. IEEE
Transactions on Neural Systems and Rehabilitation
Engineering, 14(1):55–63.
icSPORTS2014-InternationalCongressonSportSciencesResearchandTechnologySupport
176
