You are on page 1of 12

Bulletin of the JSME Vol.9, No.

1, 2022

Mechanical Engineering Journal

Vehicle’s center of gravity estimation using pavement


Weight-in-Motion
Xuan Tung PHAM*, Masaaki OKUMA*, Mulyadi BUR** and Zainal ABIDIN***
* Department of Mechanical Engineering, Tokyo Institute of Technology
2-12-1, O-okayama, Meguro-ku, Tokyo, JAPAN
E-mail: tung.phamxuan@hust.edu.vn
** Department of Mechanical Engineering, Andalas University, Padang, Indonesia.
*** Department of Mechanical and Aerospace Engineering, Institute of Technology Bandung, Indonesia.

Received: 13 September 2021; Revised: 25 October 2021; Accepted: 21 January 2022

Abstract
The center of gravity of a vehicle (CG) is a critical parameter in influencing the quality of driveability and safety,
such as for preventing rollover accidents. Therefore, it is important to identify a vehicle’s CG in practical
situations. In particular, it is strongly desired to make it possible to measure the height of the CG of trucks and
trailers freighted with various voluminous and heavy loads on business operation. The technology of Weigh-In-
Motion (WIM) is being developed to identify the gross weight of trucks and trailers and detect overload. In this
paper, the authors propose a method for the estimation of a vehicle’s CG, especially the height of CG, based on
the WIM system. The basic theory is described at first, and the validity is shown by means of basic experiments
using a light truck and a motorbike.

Keywords: Measurement, Height of center of gravity, Driveability, Gross weight of vehicle, Weigh-in-motion,
Axle load

1. Introduction

The height of a vehicle’s center of gravity (CG) is one of the most important parameters influencing the quality of
driveability and safety, such as for preventing rollover accidents. Recent static studies showed that heavy vehicles with
a higher CG are more likely to roll over compared to others (NHTSA, 2006). It also reported that the rollover of large
trucks and buses caused over 85% of fatal crashes. Thus, measuring and knowing the CG can contribute to improving
the safety of vehicles on the road. Traditionally, the height of a vehicle’s CG is often measured with the front or the rear
wheels raised a small distance by the static method using load measurement devices called “the lifted method” (Winkler
et al., 1991). This method is highly accurate. However, it is not practically effective for vehicles carrying complicated
loads or in motion. There also exist methods to estimate the CG of a vehicle in motion based on the vehicle model in roll
(Solmaz et al., 2008) or yaw motion (Albinsson et al., 2016) or using Kalman filter (Wenzel et al., 2006, Huang and
Wang, 2013, 2014). However, for the methods in these studies, it is required to know additional parameters, such as
gross weight, tire model, damper or road slope, which are difficult to be obtained for vehicles on the road.
During straight movement with acceleration or deceleration, partial amount of the normal axle load transfers from
the front to the rear axle or vice versa. This is called “transfer load.” Determining transfer load can improve the estimation
accuracy of the static axle load by WIM systems by compensating the transfer load from front to the rear wheel or vice
versa (Pham et al., 2020). The WIM considering the transfer load can be applied to estimate the height of a vehicle’s CG.
In previous studies (Pham et al., 2020, 2021), the authors presented an estimation algorithm based on an averaging
method, and basic experiments using motorbikes and a light truck were carried out to verify the theory. In the case of
non-constant speed, the acceleration rate and the height of a vehicle’s CG played an important role in the axle load
estimation. The authors proposed a method using two or more weigh-in-motion measurement devices in cascaded
placement in a traffic lane to find a vehicle’s acceleration rate (Pham et al., 2021) for improving the static axle load

Paper No.21-00322 © 2022 The Japan Society of Mechanical Engineers. This is an open access
J-STAGE Advance Publication date : 30 January, 2022 article under the terms of the Creative Commons Attribution-NonCommercial-
[DOI: 10.1299/mej.21-00322] NoDerivs license (https://creativecommons.org/licenses/by-nc-nd/4.0/).
1
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

estimation. This method improved the accuracy of acceleration measurement in comparison with the previous method
(Pham et al., 2020), that used a single weigh-in-motion measurement device. However, the heights of CG of the test
vehicles in both the studies were assumed based on the general characteristics of the vehicles.
In this paper, we propose the estimation of the CG of vehicles in motion using the WIM system. In particular, we
focus on the estimation of the height of CG because it is the most important technical challenge. The method is verified
by means of basic experiments using a light truck and a motorbike. To the best of the authors’ knowledge, there is no
method for the measurement of the height of the CG of a motorbike in running condition with a rider seated on. Therefore,
in this paper, we propose a method for it.

2. Theory of vehicle’s center of gravity estimation by Pavement Weigh-in-Motion

Pavement Weigh-in-Motion is the technology currently used to estimate the axle load and subsequently to estimate
the gross weight of vehicles on the road without enforcing them to stop. In the previous studies (Pham et al., 2020, 2021),
the authors presented the static axle load estimation based on an averaging algorithm using the plate type of measurement
devices and showed basic experiments using motorbikes and a light truck. To identify only the gross weight of vehicles,
there is no need to pay attention to the acceleration of vehicles at the time of travelling on the measurement devices.
However, the front and the rear axle loads of vehicles travelling with acceleration or deceleration differ from those to be
measured in constant speed conditions. Because the acceleration makes the load transfer from the front axle to the rear
as illustrated in Fig.1, and the deceleration does so from the rear axle to the front. The transfer load 𝑁 is expressed as
the function with respect to the mass of the vehicle, acceleration, wheelbase, and height of CG by Eq. (1).

𝑁 = 𝑚𝑎 (1)

where 𝑎 , B, ℎ , and m are the acceleration, vehicle’s wheelbase, height of CG, and mass of the vehicle, respectively.
The front and rear axle loads, Wf and Wr, measured by WIM system are expressed by Eqs. (2) and (3), respectively.

𝑊 =𝑁 −𝑁 (2)

𝑊 =𝑁 +𝑁 (3)

where Nf and Nr are the static axle loads of the vehicle’s front and rear axles to be estimated by WIM in a constant speed
condition. The mass of the vehicle is estimated by 𝑚 = 𝑊 + 𝑊 /𝑔 where 𝑔 is the gravitational acceleration.
The height of CG can be estimated by the computation using the front and rear axle loads and the acceleration of the
vehicle measured by WIM system. Consider the estimation of the height of CG of a vehicle composed of a single front
axle and rear axle using n load measurement devices placed in cascade in a traffic lane (See an example in Fig.2). The
( ) ( )
front and the rear axle load estimated with the sensor signal from the k-th measurement device, 𝑊 and 𝑊 , are
expressed by Eq. (4).

( ) ( ) ( )
𝑊 =𝑁 −𝑁
(𝑘 = 1~𝑛) (4)
( ) ( ) ( )
𝑊 =𝑁 +𝑁

In addition, the acceleration of the vehicle at travelling on the k-th measurement device, which may be measured in the
operation of WIM or using another measurement system such as optic-microwave mixing speedometer, is expressed by
𝑎( ) . Equations (5) and (6) are obtained by the combination of Eqs. (1) and (4).

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
𝑊 −𝑊 = 𝑁 −𝑁 − 𝑁 −𝑁 =𝑁 −𝑁 = 𝑚𝑎( ) − 𝑚𝑎( ) (5)

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


2
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

𝑎: acceleration
B: wheelbase
bc: longitudinal position of the center of gravity in the forward direction from the center of the
rear wheel
G: center of gravity
𝑔: gravitational acceleration
hG: height of center of gravity
m: mass of vehicle
W: axle load (the subscripts r and f mean rear and front, respectively)
Nt: transfer load appears in case of 𝑎 ≠ 0

Fig. 1 Free-Body-Diagram of 2D model of vehicle in accelerative movement,

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ℎ ℎ
𝑊 −𝑊 = 𝑁 +𝑁 − 𝑁 +𝑁 =𝑁 −𝑁 = 𝑚𝑎( ) − 𝑚𝑎( ) (6)
𝐵 𝐵

where the superscripts i and j mean the i-th and j-th measurement devices, respectively. The height of CG can be

estimated from Eqs. (5) and (6) under the assumption of 𝑎( ) ≠ 𝑎( )


as expressed by Eqs. (7) and (8).
( ) ( )
𝑊 −𝑊 𝐵
ℎ = (7)
(𝑎( ) − 𝑎( ) )𝑚

( )

( ) (8)
ℎ = ( ( )− ( ))

To reduce random error on the estimation, the height of CG can be estimated by averaging the Eqs. (7) and (8)
obtained under all combinations of the load measurement device number indices (i) and (j) (i= 1~n, j=1~n). Note here
that the numbers 1 or 2 will be practically substituted for n. For example, in the case of using two measurement devices
(n=2), it can be calculated by Eq. (9).

( ) ( ) ( ) ( )
− − −
ℎ = ( )− ( ))
(9)
(

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


23
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

where 𝑎( ) ≠ 𝑎( ) is assumed. Note that the wheelbase is assumed to be a known constant from the specification of the
vehicle.
The horizontal position of the vehicle’s CG is estimated as follows. At first, the longitudinal position, 𝑏 , in the
forward direction from the center of the rear wheel is estimated using the acceleration 𝑎( ) and the estimated front and
( ) ( )
rear axle loads 𝑊 and 𝑊 as expressed by Eqs. (10) and (11).

( )
( )
𝑏 = ( ) ( ) 𝐵+ ℎ (10)
+

( ) ( )
𝑎( ) 𝑊 − 𝑎( ) 𝑊 (11)
𝑏 = 𝐵
(𝑎( ) − 𝑎( ) )𝑚𝑔

Equations (10) and (11) are derived by means of integrating the following two force balance equations, Eqs. (12) and
(13), which are obtained according to the free-body diagram in Fig.1. The translational force balance equation in the
vertical direction is obtained as Eq. (12).

( ) ( )
𝑊 +𝑊 = 𝑚𝑔 (12)

The balance equation of the moment of force about the contact point of the rear wheel with the road surface is obtained
as Eq. (13).

( )
𝑊 𝐵 + 𝑚𝑎( ) ℎ = 𝑏 𝑚𝑔 (13)

Some load measurement devices are placed in a cascaded layout in a traffic lane to measure the right
and left wheel loads of each vehicle multiple times in a series.

Fig. 2 An example of WIM system layouts to be proposed for estimating the position of vehicles’ center of gravity

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


24
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

Travelling direction

Measurement device

(a) Placement of the load measurement device (b) Test truck moving with its left wheels travelling on
in the test lane the load measurement device in the test lane
Fig. 3 Overview of basic experiment using a light truck

Consider the position as “longitudinal position” below. First, if the front and rear axle load measurement of a vehicle
at a constant speed is possible, 𝑏 can be estimated simply using only the first term on the right-hand side of Eq. (10).
In the case that WIM measurement is conducted for a vehicle travelling with acceleration, 𝑏 can be derived from Eq.
(11) by substituting the value of ℎ estimated by Eqs. (7) and (8) in general or by Eq. (9) in case of n=2.
Second, the horizontal position of CG in the lateral direction is estimated using the balance between the right and
left side wheel loads, which can be measured using the WIM system by embedding two load measurement devices in
parallel in a traffic lane as illustrated in Fig. 2. The lateral position of CG, 𝑙 , from the geometric center line of vehicle
in the longitudinal direction, is calculated by Eq. (14).

𝑙 = 𝑊 −𝑊 + (𝑊 −𝑊 ) (14)

where 𝑇 , 𝑇 , 𝑊 , 𝑊 , 𝑊 and 𝑊 are the tread of front wheels, the tread of rear wheels, and the measured
load of front-right wheel, front-left wheel, rear-right wheel and rear-left wheel, respectively. In this paper, the wheelbase,
B in Fig.1, and the tread of wheels, 𝑇 and 𝑇 in Eq. (14), are assumed as known constants. The current intelligent
transportation systems (ITS) have the database of the specification of all officially certificated vehicles to move on public
road generally in the world. ITS can identify the detailed specifications of vehicles, such as vehicle manufacturer, year
of manufacture, vehicle classification, body identification number, body size, gross weight, number of wheel axles, each
axle load, wheelbase, the tread of wheels, seating capacity, maximum allowable load capacity, etc., almost in real time
according to visual sensing of vehicle registration plates (Won, 2020, Gholamhosseinian et al., 2021).

3. Basic experiments and results

Basic experimental studies are conducted in two cases using a light truck and a motorbike to investigate the validity
of the proposed method. The first experiment is for the estimation of the height of the CG of the light truck shown in
Fig. 3. The second one is for a motorbike, as shown in Fig. 6.

3.1 Experiment using a light truck


3.1.1 Experimental setup
The test road is constructed by embedding a load measurement device for the test light truck. Figure 3 shows an
overview of the experiment for measuring the left side wheel loads of the truck, the wheelbase of which is 1.905 meters.
The signal sampling speed is 2 kHz.
To evaluate the accuracy of the estimation, the height of CG of the light truck, including the same test driver on the
driver’s seat, is measured by an alternative method based on the gravity-dependent suspension modelling algorithm
(Kloepper et al., 2009), which measures all parameters of mass properties (mass, position of center of gravity, and inertia

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


25
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

tensor). This measurement is commissioned to Resonic Japan (Resonic, 2021). The height of CG is measured as 0.575m
by the method.

3.1.2 Result of the truck’s CG


Experiments are conducted to measure the wheel loads on each side in acceleration and deceleration conditions.
Figures 4 and 5 show the load sensor outputs of two trials of measuring the left and the right wheels with acceleration
and deceleration, respectively.

The blue and the red lines depict sensor outputs in acceleration and deceleration, respectively.
Fig. 4 Example of load sensor outputs in acceleration and deceleration conditions (left wheels).

Trial A3
300 Trial D2

250

200

150

100

50

-50
1.4 1.6 1.8 2 2.2 2.4
Time (s)
The blue and the red lines depict sensor outputs in acceleration and deceleration, respectively.
Fig. 5 Example of sensor outputs in acceleration and deceleration conditions (right wheels).

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


26
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

Table 1 Estimated results of wheel loads of the truck in acceleration condition (left side)
Velocity Acc. Gross weight
Test No. Front load (kgf) Rear load (kgf)
(km/h) (m/s2) (kgf)
Trial A1 22.3 1.3 193.1 209.9 403.0
Trial A2 23.3 2.0 195.9 225.1 421.0
Trial A3 23.8 2.2 190.5 216.4 406.9
Trial A4 24.4 1.8 193.3 219.9 413.1
Trial A5 30.3 1.4 197.9 219.2 417.1

Table 2 Estimated results of wheel loads of the truck in deceleration condition (left side)
Velocity Acc. Gross weight
Test No. Front load (kgf) Rear load (kgf)
(km/h) (m/s2) (kgf)
Trial D1 17.5 -2.4 239.2 182.3 421.5
Trial D2 18.0 -1.7 233.5 182.0 415.5
Trial D3 18.2 -2.2 241.2 177.6 418.8
Trial D4 23.4 -2.1 230.2 181.4 411.7

Table 3 The height of CG of the truck estimated from left-side wheel load measurement
(Each result is obtained by Eq. (9) using a pair of data in acceleration in Table 1 and deceleration in Table 2)

Trial A1 Trial A2 Trial A3 Trial A4 Trial A5

Trial D1 0.45 0.43 0.40 0.45 0.45


Trial D2 0.52 0.48 0.45 0.50 0.52
Trial D3 0.52 0.49 0.46 0.51 0.53
Trial D4 0.44 0.42 0.39 0.44 0.45
(meter)

Tables 1 and 2 list the estimation results of the left wheel load in acceleration and deceleration conditions. Table 3
lists the height of the truck’s CG estimated subsequently from the pairs of the acceleration and deceleration experiments.
Tables 4 and 5 list the estimation results of the right wheel load in acceleration and deceleration conditions. Table 6
lists the height of CG estimated subsequently by the pairs of the estimated wheel loads.
The mean values of the estimations of the height of the vehicle’s CG are 0.47 m with a standard deviation of 0.04
and 0.50m with a standard deviation of 0.07 from the left-side data and right-side data, respectively. As mentioned earlier,
the measurement result by Resonic Japan is 0.575 m. Therefore, it can be said that the estimation results in Tables 3 and
6 are acceptable.
The variances of the results in Tables 3 and 6 can be explained from the composition of Eqs. (7), (8) and (9). The
numerators and the denominators are composed of the difference values of estimated wheel loads and accelerations,
respectively. The difference values tend to have bigger errors than individual wheel loads and accelerations,
mathematically. The sensitivity of the acceleration to the equations is nonlinear, and it is more influential in introducing
errors than the load in the numerator. The variances of the output of the Eqs. (7), (8) and (9) tend to become bigger than
those of individual wheel loads and accelerations.
Table 7 presents the longitudinal positions of CG in the forward direction from the center of the rear wheels, which
are calculated from the results of the left-side wheel loads in acceleration and deceleration conditions. Prior to the
experiments, four wheel loads are measured in the stationary condition using a scale on-site. The result of the stationary
measurement is listed in Table 8. The longitudinal position of CG is calculated as 0.973 meters from the measurement
data in Table 8. The longitudinal position of CG is estimated accurately.

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


27
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

Table 4 Estimated results of wheel loads of the truck in acceleration condition (right side)

Velocity Acc. Gross weight


Test No. Front load (kgf) Rear load (kgf)
(km/h) (m/s2) (kgf)
Trial A1 19.0 1.2 205.9 245.5 451.4
Trial A2 19.6 1.3 203.3 246.3 449.6
Trial A3 19.9 1.2 205.0 254.7 459.7
Trial A4 24.4 1.8 208.0 247.7 455.7

Table 5 Estimated results of wheel loads of the truck in deceleration condition (right side)
Velocity Acc. Gross weight
Test No. Front load (kgf) Rear load (kgf)
(km/h) (m/s2) (kgf)
Trial D1 27.0 -2.3 243.7 204.1 447.7
Trial D2 27.4 -1.3 232.4 206.3 438.6

Table 6 The height of CG of the truck estimated from right-side wheel load measurement
(Each result is obtained by Eq. (9) using a pair of data in acceleration in Table 4 and deceleration in Table 5)
Trial A1 Trial A2 Trial A3 Trial A4
Trial D1 0.47 0.48 0.54 0.40
Trial D2 0.53 0.54 0.63 0.43
(meter)

Table 7 The estimated results of longitudinal position of CG


in the forward direction from the center of the rear wheels for truck
(Note: Each result is calculated by Eq. (11) using a pair of data in Tables 1 and 2)

Trial A1 Trial A2 Trial A3 Trial A4 Trial A5

Trial D1 0.96 0.97 0.98 0.97 0.97


Trial D2 0.98 0.98 0.99 0.98 0.98
Trial D3 0.98 0.98 0.99 0.98 0.98
Trial D4 0.97 0.97 0.98 0.97 0.97
(meter)

Table 8 Static measurement of four wheel loads of the truck including a seated driver
Left wheel load (kgf) Right wheel load (kgf)
Front 218 230
Rear 202 227
Note: the wheelbase is 1.905 m

3.2 Result of a motorbike’s CG


3.2.1 Experimental setup
The experimental setup for the measurement of a motorbike is shown in Fig. 6. The setup is the same as that used in
the experimental study reported in the paper (Pham et al., 2021).

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


28
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

Two measurement
devices

Two load measurement devices are placed in a cascaded layout in the test lane.
Wheel loads and acceleration rate are estimated (measured) using the sensor signals
in the load measurement devices. Refer the paper (Pham et al., 2021) for more detail.
Fig. 6 Overview of the experimental setup for motorbike

3.2.2 Result
First, the motorbike is run at a constant speed on two measurement devices placed on the lane. The load signal output
is obtained as shown in Fig. 7, in which the blue and red lines are the load signal outputs from the first and the second
measurement device, respectively. The front and rear wheel loads are estimated as 70 kgf and 136 kgf, respectively.
Consequently, the gross weight of the motorbike including the rider’s weight is estimated as 206 kgf (=70 kgf+136 kgf).
The wheelbase is 1.315 m, as per the specifications of the motorbike. Therefore, the longitudinal position of CG is
estimated at 0.447 m in the forward direction from the rear wheel center.
Second, the motorbike is run in an acceleration condition. The load signal output is obtained as shown in Fig. 8. The
front and the rear wheel load are estimated as 56 kgf and 148 kgf, respectively. The acceleration rate is estimated as
1.54m/s2 from the signal time delay of the rear wheel load after the front wheel and that of the second device, after the
first one. Therefore, the height of CG is estimated as 0.547 m.
To evaluate the accuracy of the estimation result (0.547 m), we conduct a measurement of the height of the CG by
another method provided by Resonic Japan (URL of Resonic Japan) as same for the light truck in Section 3.1. Figure 9
shows an overview of the measurement at Resonic. The measurement result is 0.646 m. It can be said that the difference
between two results is acceptable.

150

100

50

1.5 2 2.5
Time (s)
Blue and red lines depict the sensor signals from the first device and the second one
that the test motorbike traveled on, respectively.
Fig. 7 Sensor outputs from two measurement devices in a constant speed

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


29
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

200

150

100

50

1.4 1.6 1.8 2 2.2


Time (s)
Blue and red lines depict the sensor signals from the first device and the second one
that the test motorbike traveled on, respectively.
Fig. 8 Sensor outputs from to measurement devices in an acceleration condition

Rider’s head and


body are fixed with
the supporting frame

Test motorbike
tightly fixed with
the frame structure

Supporting frame
structure

Platform

Springs

This measurement can obtain all mass properties (mass, CG, inertial tensor) of an object. The
upper part on the springs is set up to behave as a rigid body in free vibration with small
displacement (max displacement at any points in the upper structure is controlled to be less than
approximately 1 cm) at low freqeuencies ( less than 1 Hz) for measurement. The mass properties
of the upper part can be estimated by measuring the frequencies of the free vibration. The mass
properties of the test motorbike including the rider can be obtained by extracting those of the
frame structures from those of the total upper part.
Fig. 9 An overview of the setup to measure CG height of test motorbike
as a referential experiment carried out by Resonic Japan

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


10
2
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

4. Conclusions

In this paper, the authors presented a novel method to estimate the height of the CG of vehicles by using a pavement
WIM system and demonstrated that the method is promising for practical estimation in the future. Basic experimental
studies are conducted using a light truck and a motorbike. The mean value of the estimations of the height of the light
truck’s CG from the left-side and the right-side data are 0.47 m and 0.50 m, respectively. The height of the motorbike’s
CG is estimated as 0.547 m.
The accuracies of the results are verified by comparison with the results obtained by a stationary type of measurement.
The measurement results of the light truck and the motorbike by the method are 0.575 m and 0.646 m, respectively. Note
here that the stationary type of measurement is a commercialized unique method to measure the mass properties (mass,
CG and inertial tensor) of various objects such as mechanical components and machines.
It can be said that the difference between the results by two methods is acceptable. The improvement to be made for
more accurate measurement of acceleration in future works exists. The improvement will contribute to enhance the
estimation accuracy and reliability of the method proposed in this paper.
The function of estimating a vehicle’s CG position will expand the utility worth of WIM systems. WIM systems
with the function will contribute to traffic safety in the future by being embedded at various places, for detecting overload
trucks on road and safety checks at tollgates and freight terminals at harbors and railway stations. Therefore, integrated
computation with the output data (axle load and the position of CG, especially its height) from WIM and a vehicle’s
navigation system can contribute significantly to autonomous vehicle technology and transportation safety management
as an advanced intelligent transportation system in the future.

Acknowledgments

The authors are grateful to Dr. T. Kawaguchi, the president of Resonic Japan Ltd., for his support, and Assoc. Prof. H.
Sakamoto, of the Tokyo Institute of Technology, and Dr. Meifal Rusli and Dr. Lovely Son, lecturers at Andalas University,
for their valuable discussions.

References

Albinsson A., Bruzelius F., Pettersson P., Jonasson M. and Jacobson B., Estimation of the inertial parameters of vehicles
with electric propulsion. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile
Engineering, 230, 9, (2016), pp. 1155–1172.
Gholamhosseinian A, Seitz J., Vehicle Classification in Intelligent Transport Systems: An Overview, Methods and
Software Perspective, IEEE Open Journal of Intelligent Transportation Systems, vol. 2, (2021), pp. 173-194.
Huang X. and Wang J., Center of gravity height real-time estimation for lightweight vehicles using tire instant effective
radius, Control Engineering Practice, 21, 4, (2013), pp. 370–380.
Huang X. and Wang J., Real-time estimation of center of gravity position for lightweight vehicles using combined AKF-
EKF method. IEEE Transactions on Vehicular Technology, 63, 9, (2014), pp. 4221–4231.
Kloepper R. A measurement system for rigid body properties enabled by gravity-dependent suspension modelling,
Dissertation, Tokyo Institute of Technology, (2009).
Kloepper R. and Okuma M., Experimental identification of rigid body inertia properties using single-rotor unbalance
excitation. Proceedings of the Institution of Mechanical Engineers, Part K: Journal of Multi-Body Dynamics, 223,
4, (2009), pp. 293–308.
National Highway Traffic Safety Administration (NHTSA), Traffic Safety Facts 2004. A Compilation of Motor Vehicle
Crash Data from the Fatality Analysis Reporting System and the General Estimates System, Technical Report,
(2006).
Pham X.T., Okuma M., Bur M. and Abidin Z., An estimation method for pavement weigh-in-motion system with
preliminary experiment. Mechanical Engineering Journal, Vol. 7, No, 6 (2020), DOI: 10.1299/mej.20-00405.
Pham X.T., Okuma M., Bur M. and Abidin Z., Basic study on weigh-in-motion of vehicles in acceleration and
deceleration, JSME Mechanical Engineering. Journal, Journal Vol.8, No. 1 (2021), DOI: 10.1299/mej.19-00368.
Resonic GmbH, Resonic inertia measurement, available from <http://www.resonic.jp/index.php>, (accessed on 6

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


11
2
Pham, Okuma, Bur and Abidin, Mechanical Engineering Journal, Vol.9, No.1 (2022)

September, 2021).
Solmaz S., Akar M., Shorten R. and Kalkkuhl J., Real-time multiple model estimation of centre of gravity position in
automotive vehicles. Vehicle System Dynamics, 46, 9, (2008), pp. 763–788.
Wenzel T.A., Burnham K.J., Blundell M.V. and Williams R.A., Dual extended Kalman filter for vehicle state and
parameter estimation. Vehicle System Dynamics, 44, 2, (2006), pp. 153–171.
Winkler C.B., Campbell K.L., Mink C.E. and Center of Gravity, (1991) Height: A Round-Robin Measurement Program.
Won M., Intelligent Traffic Monitoring Systems for Vehicle Classification: A Survey, IEEE Access, vol. 8, (2020), pp.
73340-73358.

[DOI: 10.1299/mej.21-00322] © 2022 The Japan Society of Mechanical Engineers


12
2

You might also like