Materials Today: Proceedings: Harold Villacís Jara, Iván Zambrano Orejuela, J.M. Baydal-Bertomeu

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Materials Today: Proceedings xxx (xxxx) xxx

Contents lists available at ScienceDirect

Materials Today: Proceedings


journal homepage: www.elsevier.com/locate/matpr

Study of the ergonomic risk in operators of an assembly line using the


RULA method in real working conditions through the application of a
commercial sensor
Harold Villacís Jara a,⇑, Iván Zambrano Orejuela b, J.M. Baydal-Bertomeu c
a
Faculty of Engineering, Universidad de La Sabana, Chía, Colombia
b
Faculty of Mechanical Engineering, Escuela Politécnica Nacional, Quito, Ecuador
c
Instituto de Biomecánica de Valencia, Universidad Politécnica de Valencia, Valencia, Spain

a r t i c l e i n f o a b s t r a c t

Article history: Assessing ergonomic risk in real working conditions is a challenge for ergonomists, who use observa-
Available online xxxx tional methods to collect information, with RULA (Rapid Upper Limb Assessment) being one of the most
widely used. The disadvantage of these methods is that data collection is subjective and is based on esti-
Keywords: mates of projected angles on photos and/or videos, which leads to variations according to the criteria of
RULA each evaluator. For this reason, this work proposes the use of the Kinect V2 commercial sensor, as a tool
Kinect V2 for ergonomic assessment in real working conditions in industry. In the study, data such as planes, angles
Ergonomic risk
and positions of the worker were obtained, which the RULA method needs to determine the existence of
Ergonomic assessment
Real working conditions
ergonomic risk. The application of the sensor was carried out in real working conditions and it was per-
Marker-less motion capture system formed to an operator of a bridge beam assembly line in a metalworking factory. Then, the trajectory and
angle curves were obtained by means of an algorithm; and were compared with those of a standard
motion capture system in a photogrammetry laboratory. Finally, the RULA score obtained by the imple-
mented system was validated by an expert.
Ó 2022 Elsevier Ltd. All rights reserved.
Selection and peer-review under responsibility of the scientific committee of the Advances in Mechanical
Engineering Trends

1. Introduction Vignais et al. [6] show a study of application of inertial sensors


on the upper body to assess ergonomic risk with the RULA method.
The risk of musculoskeletal disorders in workers is related to However, these systems have several disadvantages, one of which
their posture and movements in the workplace [1]. There are is that they require the placement of sensors in the worker’s body,
several methods, mostly observational, to assess ergonomic risk which prevents them from performing their tasks normally [2].
in the workplace, such as: RULA, REBA and OWAS [2]. Nowadays, Another disadvantage is that these sensors can be disturbed if
studies are conducted in the industry to assess the risk in work- applied in an electromagnetic environment, so measurements
ers associated with a high postural load, which can cause disor- would not be reliable [3]. In addition, in the case of optoelectronic
ders in the upper limbs of the body. In this research, the RULA motion analysis systems, they are very expensive, so they are only
(Rapid Upper Limb Assessment) method is used, which involves applied under laboratory conditions, but not in real working
the collection of information that is usually obtained by subjec- conditions.
tive observation or a simple estimation of angles projected in This is where the importance of this study lies, in which the use
photos and videos [3]. This makes the inaccuracy and variability of a low-cost commercial sensor (Kinect V2) was proposed in the
from one observer to another quite considerable, as described in ergonomic risk assessment in real working conditions. In this
their studies Burdorf and Laan [4] and Fagarasanu and Kumar research the application of the commercial sensor for ergonomic
[5]. assessment was carried out to an operator of a bridge beam assem-
bly line in a metalworking company.

⇑ Corresponding author.

https://doi.org/10.1016/j.matpr.2021.07.482
2214-7853/Ó 2022 Elsevier Ltd. All rights reserved.
Selection and peer-review under responsibility of the scientific committee of the Advances in Mechanical Engineering Trends

Please cite this article as: H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu, Study of the ergonomic risk in operators of an assembly line
using the RULA method in real working conditions through the application of a commercial sensor, Materials Today: Proceedings, https://doi.org/10.1016/
j.matpr.2021.07.482
H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu Materials Today: Proceedings xxx (xxxx) xxx

1.1. State of the Art principle of adaptation of work to the person, which is the objec-
tive of occupational ergonomics.
One of the first investigations of the Kinect V1 for its application
in ergonomics is presented in [7], published in 2011, showing a 2.3. The RULA method
comparison of the data from this sensor with a pattern motion cap-
ture system, which provided good results. Then, in 2014, Microsoft Within ergonomic risk identification methods, the RULA
included a new sensor model, the Kinect V2, which has faster method is one of the most widely used observational posture
response and better accuracy than the previous version. Regarding assessment tools, which was developed for use in ergonomic work-
ergonomics, published in 2016 [3], a study with the application of place research where disorders of upper body limbs related to pos-
the Kinect V2 in real working conditions is described, in which it is tures and loads occur [9]. Given a RULA posture, a score will be
indicated that the RULA scores obtained with the help of the sensor obtained from which a certain level of performance is established.
and by ergonomics experts, had similarity in about 75% of cases This level of performance will indicate whether the posture is
and suggests a deeper analysis about the location of the sensor acceptable or to what extent changes or redesigns in the position
to decrease the error. Then, published in 2017 [8], a study shows are necessary.
an ergonomics assessment for tractor drivers, suggesting not only RULA divides the body into two groups, Group A which includes
the identification of risk postures, but a continuous real-time anal- the upper limbs (arms, forearms and wrists) and Group B, compris-
ysis with the RULA method. ing legs, trunk and neck [9]. The final RULA score ranges from 1 to 7
As explained above, to date there is little research into the and proposes the so-called levels of performance on the work place,
application of such commercial sensors in the field of ergonomics. which are detailed below:
Score 1 or 2 (level of performance 1): The worker’s task presents
an acceptable risk and no changes are required.
2. Theoretical framework
Score 3 or 4 (level of performance 2): The worker’s task needs
further study, changes may be required.
As explained in the State of the Art, the assessment of the
Score 5 or 6 (level of performance 3): The task needs changes; the
potential risks of musculoskeletal disorders (MSD) under real
job redesign is suggested.
working conditions has its complexity because the specialist must
Score 7 (level of performance 4): The task needs urgent changes.
determine angles and real positions of critical postures on pho-
tographs. At present, several methods have been developed to
assist in ergonomic risk assessment, which can be divided into 3. The material and method
three groups, according to measurement technique: self-
assessment methods, direct measurements and observational This section describes the elaboration of a tool to assist ergono-
methods [3]. mists or industrial and occupational safety personnel of companies
Self-assessment methods have the disadvantage of being quite in the assessment of the ergonomic risk of workers in the industrial
subjective, as they are based on questionnaires, checklists and field, using the RULA method. This tool consists of a pseudo-code
interviews that can lead to biased interpretations. Direct methods, and an interface (both made in Matlab). In the interface, the user
on the other hand, collect information from a system with sensors manually enters parameters that are out of range of the sensor.
placed in the worker’s body, but that is uncomfortable, costly and Also, the interface indicates the RULA score and the level of perfor-
virtually impossible to implement in real working conditions [2]. mance of the person analyzed in real time.
Regarding observational methods, they are widely applied in the
industry and consist of direct observation to the worker during 3.1. Data acquisition and skeleton tracking
the performance of his activities. However, these methods have
the main disadvantage of requiring a field expert to perform a Kinect V2 is a system that provides video images of three types:
time-consuming posture analysis, more than their veracity is RGB, depth, and infrared. With this, the first step is to generate an
related to the quality of information collected and specialist object in Matlab by specifying the type of image required from the
estimates. sensor, that is, RGB and depth.
The data is organized into an array of matrices, each array con-
sists of the position data organized into 3x25 matrices with the X,
2.1. BTS SMART D motion analysis system
Y, Z data of the 25 points detected in space (Fig. 1).

The biomechanical motion analysis system used for the valida-


3.2. Data processing
tion of the Kinect V2 is BTS SMART D of BTS Bioengineering, within
the company BioFoot, in Quito – Ecuador. This is a high-precision
Once the connection between the sensor and the computer is
optoelectronic system for motion analysis in the clinical, sporting
established and the skeleton tracking data is verified, the anatom-
and industrial fields.
ical planes for subsequent projections and angular calculations are
obtained. The calculation process begins with the definition of the
2.2. Musculoskeletal disorders following planes (Fig. 2):

Jobs that involve prolonged efforts or repetitive movements  Trunk plane: determined through the spine and the right and
increase risk factors in the employee’s health. For this reason, left hip, i.e. it consists of points 13, 17, 21 (Fig. 1).
physical stress accumulates in the worker and in the medium or  Sagittal plane: is perpendicular to the straight line that connects
long term, health problems appear, mainly musculoskeletal disor- the right hip with the left hip and passes through the center of
ders. Musculoskeletal disorders are a set of inflammatory or degen- the hip, that is, the plane is formed by points 13, 17, 1 in Fig. 1.
erative lesions of bones, muscles, tendons, joints, ligaments, blood  Front plane: is a vertical plane that passes through the right hip
vessels and nerves. and left hip, i.e., formed by points 13, 17 in Fig. 1.
There is no single solution to avoid such problems, then the  Cross plane: is a horizontal plane that passes through the right
general way of preventing ergonomic risk is to apply the general and left hips.
2
H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu Materials Today: Proceedings xxx (xxxx) xxx

aðx  x1Þ þ bðy  y1Þ þ cðz  z1Þ ¼ 0 ð1Þ

Ax þ By þ Cz þ D ¼ 0 ð2Þ

3.2.2. Calculation of angles between joints


Some of the required angles can be obtained directly with one
equation; however, there are other angles that need additional
! !
processes. The angle between two vectors A and B can be deter-
! ! ! !
mined from the scalar product A  B ¼ k A kk B kcosðhÞ indicated
in (3).
! !
A B
h ¼ arccos !! ð3Þ
AB
Another important aspect is to consider how a vector is pro-
jected onto a plane in space. The formula for obtaining the projec-
!
tion of the vector A on a three-dimensional plane that has a
!
normal B is shown in (4). The equation to obtain the perpendicular
!
component of the projection of a vector A on a 3-D plane.
Fig 1. Points detected in the body by Kinect V2 with Kin2 Toolbox. 1) Coxis. 2) Half ! !
of spinal column. 3) Neck. 4) Head. 5) Left shoulder. 6) Left elbow. 7) Left wrist. 8)
ƒƒ! ! A B
Left hand. 9) Right shoulder. 10) Right elbow. 11) Right wrist. 12) Right hand. 13) AkB ¼ B  ð4Þ
Left hip. 14) Left knee. 15) Left ankle. 16) Left foot. 17) Right hip. 18) Right knee. 19) !2
B
Right ankle. 20) Right foot. 21) Spine. 22) Middle left finger. 23) Left thumb. 24)
Middle right finger. 25) Right thumb. !
ƒƒƒ! B
A?B¼AB 2 ð5Þ
!
B
The RULA method starts by rating the position of the arm [9].
The angle of flexion/extension of the arm, was calculated between
the vector projection of the arm on the sagittal plane and the trunk
plane. To determine the abduction of the arms, the angle between
the vector spine-shoulder and the vector shoulder-elbow was calcu-
lated. A threshold of 150 degrees was set, in which, if the angle is
greater than this value, the arm is considered abducted.
For the elevation of the shoulders the angle between the vector
spine-neck and vector spine-shoulder was taken into account. A
threshold of 100 degrees was set, i.e., if the angle is less than this
value, the shoulder is considered elevated.
As for the forearm, the angle between vectors shoulder-elbow
vector and wrist-elbow was calculated. To determine whether the
forearm crosses the midline of the body or it is on the side of the
body, an auxiliary plane called shoulder plane that contains left
and right shoulders was first calculated, and is perpendicular to
the trunk plane. The relative positions of the wrist projections on
this auxiliary plane were then analyzed.
Regarding the wrist score, the angle formed by vector middle
finger-wrist and vector elbow-wrist was considered.
As far as the angle of the neck is concerned, it was calculated
between vector head-neck and the trunk axis. The lateral inclina-
tion angle of the neck was calculated between vector spine-
shoulder vector and spine-neck vector, if the angle is between 110
Fig 2. Anatomical planes. degrees and 115 degrees it was considered that there is no lateral
inclination of the head.
The evaluation also considers the angle of the trunk, which was
calculated between the normal vector of the trunk plane and the
3.2.1. Planes calculation normal vector of the front plane. To determine the rotation of
For the calculation of a plane, the three points that form it must the trunk, an auxiliary plane containing the two shoulders and
be considered. Thus, assuming three joints with coordinates A (x1, spine was obtained (auxiliary shoulder plane). Then, the angle
ƒ! ƒ! between the normal vector of this auxiliary plane and the normal
y1,z1), B (x2,y2,z2) and C (x3,y3,z3), the vectors and AB and AC
ƒ! ƒ! vector of the front plane was calculated. A threshold of 10 degrees
are obtained. Then the vector product of AB and AC is the normal was established to consider that there is rotation of the trunk.
direction of the desired plane; therefore, the plane that contains For the lateral inclination of the trunk, the angle between the
!
the point (x1,y1,z1) and has the normal vector n ¼ ða; b; cÞ is vector that joins the hips and the vector that joins the shoulders
obtained with (1). Also, for calculation purposes, (1) can be repre- was calculated. Theoretically, if a person is standing, this angle
sented as (2), which is the general or implicit equation of the plane. should approach 0 degrees. Therefore, after testing with the sensor
3
H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu Materials Today: Proceedings xxx (xxxx) xxx

it was possible to determine an appropriate threshold, thus if the


angle is greater than 10 degrees the trunk is inclined.
Moreover, the position of the feet in the vertical coordinate was
considered for the leg score. After tests with different sensor posi-
tions, it was established that, if there is a difference greater than
7 cm, the person is supported on one foot.
Although the Kinect V2 has many improvements over its first
version, it has limitations to measure certain parameters such as:
neck rotation, wrist rotation and wrist deviation. Therefore, these
will be entered into the manually developed interface. Also,
parameters such as loads and type of activity, which are out of
range of Kinect V2, will be entered in the same way.

3.2.3. Interface features


As seen in Fig. 3, on the left side of the GUI all manual input
parameters are configured as: wrist deviations, wrist rotation, head
rotation, activity type and supported load. Also, for the position of
the legs, automatic or manual operation can be chosed. If configu-
ration is automatic, the score is given according to the calculation
that was previously explained, and if configuration is manual, it
can be selected whether the person is with both feet or one sup-
ported only. The latter was implemented because on many occa-
sions there is occlusion of the working environment in the lower Fig 4. Location of the Kinect V2 relative to the operator’s front line. The following
members of the body. In addition, the RULA assessment can be per- angular positions are presented for the Kinect V2: I) 30 degrees. II) 60 degrees. III)
90 degrees. IV) 120 degrees. The one chosen for analysis is position II.
formed over a given time interval; that is, scores of each frame can
be saved in order to have a RULA score with different percentages
of level of performance. and unloading interval. In the position of Kinect V2 at 60 degrees
of the frontline (position II of Fig. 4), it was where the best skeleton
3.3. Application of the sensor in the industry tracking results were obtained.
The algorithm calculated the RULA score in real time and
The task chosen to perform the analysis was specifically in a records the score at defined intervals and then, averages of the
hydraulic shear, where the operator, changes the cut material entire analyzed task were obtained.
without lifting tooling, at intervals lasting approximately 13 s
(loading and unloading), for 2 to 3 min and with breaks of 15 to
20 min between cutting lots. 3.4. Description of the analyzed task
Kinect V2 was placed 3 m apart from the operator, at a height of
1.2 m above the ground and varying 4 angular positions (30, 60, 90 The task lasted approximately 13 s and the average weight
and 120 degrees) from the front line of the worker (Fig. 4), avoiding lifted was 6 kg. The task consists of the following sequential move-
interrupting the operator’s free performance during the loading ments (Fig. 5):

Fig 3. Implemented graphical user interface (GUI).

4
H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu Materials Today: Proceedings xxx (xxxx) xxx

3.6. Validation of the system with an expert in occupational safety and


health

One of the specific objectives of this study is to validate the


results with those obtained by an expert of ergonomic assess-
ments; thus, an occupational safety and health professional from
Baca Ortiz Pediatric Hospital (Quito-Ecuador) was contacted to
analyze the worker’s task. To do this, the expert was provided with
videos of actual movements in the industry and simulated move-
ments in the laboratory, so that he can perform the measurements
of angle projections on the images.
Fig 5. Skeleton tracking in: (A) Collecting cut plates. (B) Lifting of cut sheet metal.
(C) Repositioning the cut sheet metal.
Consecutively, the expert identified the highest risk posture and
entered the parameters required by the RULA method into an Excel
template, which calculates the final posture score. Subsequently, a
Stage A: The operator flexes his knees and tilts the trunk for- comparison of the expert’s result was made with that obtained by
ward to take the cut sheet metal with one hand. the system implemented in this research.
Stage B: The operator rises again in its upright position and
changes the sheet metal he took to the other hand.
Stage C: The operator flexes his knees and tilts the trunk for- 4. Results and discussion
ward to take the sheet metal with one hand.
The objective of this research is to test the system in real work-
ing conditions, therefore, once the ergonomic risk task was chosen,
3.5. Validation under laboratory conditions the tests were carried out with the sensor and the algorithm imple-
mented, with which a RULA score of 6 was obtained for the left and
Once the ergonomic risk task was assessed with the system right side of the body in the most critical position of the worker.
implemented, it was necessary to check its validity under labora- This RULA number indicates a level of action 3, which means that
tory conditions. For this, the movements of the worker’s task were task redesign is needed.
replicated in a specialized laboratory with the BTS SMART motion
capture system.
4.1. Validation of the system under laboratory conditions

3.5.1. Features of the photogrammetry laboratory used BTS SMART and Kinect V2 systems recorded the trajectories of
The biomechanical motion analysis laboratory consists of 6 BTS the movement with different software, so that each data acquisi-
Bioengineering optoelectronic cameras, BTS SMART D information tion is analyzed independently. Then, it is determined that the tra-
processing unit and SMART Capture software. jectories of both the laboratory and Kinect V2 systems are very
similar and the trajectories closest to the laboratory system are
obtained on the Y axis, as shown in Fig. 7.
3.5.2. Validation protocol definition At this point, it is important to mention that the RULA method
A volunteer was recruited as a participant to simulate the oper- considers certain angles to get the final score, so the next step is to
ator’s movements in the photogrammetry lab. The first step is to calculate the values of these angles with the data acquired by the
place the markers on the body of the subject who will repeat the BTS Smart and Kinect V2 system. To determine the accuracy of
movements. The markers were placed in similar positions to the the Kinect V2 sensor, the Mean Square Error (RMSE) was obtained.
points detected by Kinect V2. Also, The Kinect sensor was then Also, the Normalized Mean Square Error (NRMSE) is calculated
placed in the same position as when applied in the industry which divides the RMSE for the range of values measured by the
(Fig. 6). marker system, as shown in Table 1.

Fig 6. Simultaneous data acquisition with BTS Smart system and Kinect V2.

5
H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu Materials Today: Proceedings xxx (xxxx) xxx

(in this case forearm-wrist-middle finger) accuracy was less (errors


between 22.87% and 54.76%). In other words, at angles captured by
the system with the Kinect sensor, which had a limited range of
motion, there was a greater error. Nevertheless, a good accuracy
was reached when capturing movement of the largest dimension
members and a wider range of motion.

4.2. Validation of the system with an expert in occupational safety and


health

The ergonomics expert performed the analysis of the task and


selected the highest risk posture of the worker to obtain the final
RULA score. In this case, the recreation of the position analyzed
by the expert is shown in Fig. 9, when the worker bended his knees
to accommodate the cut sheet metals. A score of 7 was obtained
with the professional, and 6 with the Kinect V2 system in the activ-
ity of the worker being studied. Both scores agree that changes
should be made to the task, as it represents a risk of possible mus-
culoskeletal disorders. This difference in scores is possibly due to
Fig 7. Right wrist trajectory in the Y axis obtained by the BTS Smart and Kinect V2 the different angle measurement references, as these criteria vary
system. The position is normalized. according to each evaluator. It must be said that, in the case that
the score of the implemented software differs from that obtained
by the ergonomist, the last word will always have the expert;
Table 1
RMSE and NRMSE from angles obtained with Kinect V2 and BTS Smart System.
because the software is only a tool to help the assessment.
In short, it can be said that the Kinect V2 is applicable in the
Angle RMSE NRMSE study of ergonomic assessments for the case of this task and speci-
Right forearm 156,811 0,1121 fic job. However, further study of hand movement detection is
Left forearm 147,717 0,2172 needed, where Kinect V2 significantly decreases tracking
Trunk 145,120 0,2379
capability.
Right wrist 80,051 0,2287
Left wrist 93,098 0,5476
Right knee 270,792 0,2472
Left knee 261,506 0,2577
4.3. Limitations

Although Kinect V2 can detect wrist, hand, middle finger, and


thumb, it is not feasible to determine whether the wrists are
In Fig. 8 it can be seen the angles of the right forearm, where the
deflected or rotated, as they perform relatively small movements
graphs suggest a very significant similarity of the motion capture
that are difficult to detect because of sensor sensitivity. For this
of the Kinect V2 and the marker-based system, giving an RMSE
reason, the deviations and turns of the wrist must always be
of 15,6811 and an NRMSE of 0.1121 (11.21%).
entered manually by the user of the system. Similarly, there is
According to figures showing the trajectories and angles, there
the limitation of determining the rotation of the head, since the
was a great match between the motion capture of the BTS SMART
Kinect V2 detects this part of the body and neck only as points;
system and the Kinect V2 system. However, a considerable oscilla-
therefore, the input of this parameter must be manual.
tion in both systems was observed, so the curves differ somewhat
Due to the occlusion of the environment when testing under
between the marker-based system and Kinect V2. With this, it can
real working conditions, the sensor may not be able to determine
be said that for angles between members that had a short distance
the position of the lower limbs of the body; reason why the option

Fig 8. Angle of the right forearm. Fig 9. Risk position of the task analyzed according to ergonomics expert.

6
H. Villacís Jara, Iván Zambrano Orejuela and J.M. Baydal-Bertomeu Materials Today: Proceedings xxx (xxxx) xxx

of manual filling is added in the interface, as well as the type of or by other parts of the same body. The Kinect V2 has an internal
activity and load supported. algorithm that allows to infer the joints that present occlusion.
The detection depth range according to the Kinect V2 manual is Nevertheless, in these cases the RULA score would not be reliable.
up to 4.5 m, thus the placement position of the Kinect V2 should The best results are presented when the worker is facing the cam-
vary according to the task to be analyzed until the best conditions era or with little deviation from the frontal axis, with good lighting
are found: skeleton tracking, ambient light and environmental and little ambient occlusion.
occlusion. Finally, the Kinect V2 sensor has a very acceptable reliability for
capturing motion. To use this sensor in biomechanical studies, it
4.4. Future works must be considered that it provides general information on the
movement of the body, and that it is difficult to detect small vari-
This project focuses on the field of research dealing with the use ations, that is, the Kinect V2 responds much better when the per-
of low-cost depth sensors, such as Kinect V2, in ergonomic assess- son’s movements are not very fast. Therefore, a more in-depth
ments, which has been very little explored; and therefore, there are study in this field is necessary to improve its precision.
a variety of studies and projects that can be developed on this
topic, for instance: CRediT authorship contribution statement

 Ergonomic assessment in other jobs: There are other tasks and Harold Villacís Jara: Conceptualization, Methodology, Soft-
jobs in which it is possible to explore such as: offices, vehicle ware, Investigation, Validation, Writing - original draft, Visualiza-
drivers, health care-related jobs, etc. tion, Data curation. Iván Zambrano Orejuela: Supervision,
 Multi-sensor systems: As mentioned above, one of the limita- Validation, Investigation, Writing - review & editing. J.M. Baydal-
tions of the system implemented is the occlusion, both of the Bertomeu: Supervision.
environment and of the subject of analysis. A solution to this
problem may be to perform a system with two or more sensors, Declaration of Competing Interest
in order to have a greater range of view of the worker.
 Other ergonomic assessment methods: This study aims to lead The authors declare that they have no known competing finan-
further research involving other ergonomic assessment cial interests or personal relationships that could have appeared
methods. to influence the work reported in this paper.

5. Conclusions References

In the present study, a low-cost sensor, in this case the Kinect [1] L.G. Wiedemann, R. Planinc, y M. Kampel, Ergonomic-monitoring of office
workplaces using Kinect, Lect. Notes Comput. Sci. (2014) 275–278.
V2, has been used as a useful tool in the field of ergonomics. It
[2] V.M. Manghisi, A.E. Uva, M. Fiorentino, V. Bevilacqua, G.F. Trotta, y G. Monno,
has been proven that low-cost sensors can constitute a support Real time RULA assessment using Kinect v2 sensor, Appl. Ergon 65 (2017) 481–
instrument for the evaluator, in order to have measurements closer 491.
to reality, and therefore a more complete and reliable ergonomic [3] P. Plantard, H.P.H. Shum, A.S. Le Pierres, y.F. Multon, Validation of an ergonomic
assessment method using Kinect data in real workplace conditions, Appl. Ergon.
study. 65 (2017) 562–569.
From the curves obtained, it could be verified that the motion [4] A. Burdorf, J. Laan, Comparison of methods for the assessment of postural load
capture results of Kinect V2 are not exactly the same as those of on the back, Scand. J. Work. Environ. Heal 17 (6) (1991) 425–429.
[5] M. Fagarasanu, y.S. Kumar, Measurement instruments and data collection: a
the BTS SMART marker-based system. However, very good approx- consideration of constructs and biases in ergonomics research, Int. J. Ind. Ergon
imations can be obtained in the general movement of the body. It 30 (6) (2002) 355–369.
was found that Kinect V2 gives better results when the joints have [6] N. Vignais, M. Miezal, G. Bleser, K. Mura, D. Gorecky, y.F. Marin, Innovative
system for real-time ergonomic feedback in industrial manufacturing, Appl.
a greater range of motion, and angles with better approximation Ergon 44 (4) (2013) 566–574.
are obtained when they are between adjacent members with a [7] T. Dutta, Evaluation of the KinectTM sensor for 3-D kinematic measurement in
greater length. the workplace, Appl. Ergon. 43 (4) (2012) 645–649.
[8] T. Schempp, y.S. Boettinger, Ergonomic Checks in Tractors through Motion
One of the problems that exists when conducting the study in Capturing, Chem. Eng. Trans. 58 (2017) 31–36.
working conditions is the occlusion of the environment, which [9] M. Lynn, y.N. Corlett, RULA: A survey method for the investigation of work-
means that part of the body is hidden by equipment or machinery, related upper limb disorders, Appl. Ergon 24 (2) (1993) 91–99.

You might also like