Content uploaded by Laura Gastaldi
Author content
All content in this area was uploaded by Laura Gastaldi on Mar 27, 2018
Content may be subject to copyright.
Analysis of Exoskeleton Introduction in Industrial Reality:
Main Issues and EAWS Risk Assessment
Stefania Spada
1
, Lidia Ghibaudo
1(✉)
, Silvia Gilotta
1
,
Laura Gastaldi
2
, and Maria Pia Cavatorta
2
1Fiat Chrysler Automobiles - Manufacturing Engineering – Ergonomics Torino, Turin, Italy
{stefania.spada,lidia.ghibaudo}@fcagroup.com,
silvia.gilotta@gmail.com
2Department of Mechanical and Aerospace Engineering, Corso Duca Degli Abruzzi 24,
10129 Turin, Italy
{laura.gastaldi,maria.cavatorta}@polito.it
Abstract. Exoskeletons are part of the technological and organizational inno‐
vation sought by the fourth industrial revolution to support and re-launch the
manufacturing area. In the present study, we described the experimental protocol
designed to test the usability and acceptance of an upper limbs passive exoskel‐
eton. In total, 42 workers from FCA plants volunteered to participate in the
research study. The testing campaign included static and dynamic tests aimed at
evaluating the potential benefit of the exoskeleton (lessen muscle strain, higher
comfort rating and dexterity) vs. possible restrictions to movements and work-
device interactions in tasks resembling work activities. Open questions remain
on how to assess the biomechanical workload risk, especially in the design phase,
for which holistic methods like EAWS are needed.
Keywords: Passive exoskeleton · Upper limbs · Human-robot cooperation ·
Fourth industrial revolution · Usability · Workload
1 Introduction
The fourth industrial revolution faces the challenge of European industry growth with
new policies to re-launch the manufacturing area, through technological and organiza‐
tional innovation. Technological evolution has progressed from automated systems to
human-robot collaboration, in which the human and the robot combine into one inte‐
grated system under the control of the human. The scope is to safeguard the worker’s
wellbeing while optimizing productivity and system performance.
An exoskeleton is a wearable device in which the physical contact between the
operator and the mechanical structure allows a direct exchange of mechanical power
and information signals [1–3]. The exoskeleton technology has originated from the
military [4, 5] and rehabilitation [6, 7] fields and is now rapidly growing in industrial
settings. Here exoskeletons could be useful when other preventive measures are not
feasible or effective (i) to lower worker’s fatigue, thus leading to increased worker’s
alertness, productivity and work quality, (ii) to support quality and experienced
© Springer International Publishing AG 2018
R.S. Goonetilleke and W. Karwowski (eds.), Advances in Physical Ergonomics
and Human Factors, Advances in Intelligent Systems and Computing 602,
DOI 10.1007/978-3-319-60825-9_26
personnel in the work force longer, and (iii) to reduce work related musculoskeletal
disorders [1].
Exoskeletons for industrial applications include upper limb and back supports, to
reduce the load on the shoulder and back muscles while holding awkward postures.
Chairless chairs, that can stiffen and lock in place, aim at decreasing the fatigue of
crouching or standing in the same position for an extended period, while powered
exoskeletons are designed to enhance the strength and the resistance to fatigue in
stressful jobs. These exoskeletons, either passive or active, require different approaches
towards fulfilling requirements such as usability, acceptability at the workplace and
potential safety issues [8–10]. Literature studies generally look into the variation in the
EMG level of the primary muscles involved in analyzed activity, to suggest a possible
reduction in the overall muscle work when using an exoskeleton [11–15]. However,
very few research projects and validation activities have been carried out concerning
the estimation of the biomechanical load in exoskeleton-assisted work tasks and the
potential impact of exoskeletons for the risk assessment and the work methods. In addi‐
tion, evaluation studies on existing exoskeleton prototypes are often limited to laboratory
trials on university students or staff [16–19]. Only in [12], a passive lift-assist device is
tested in an automotive manufacturing plant with operators.
Testing protocols usually involve a repeated measure type experimental design,
including within-subject comparisons of without and with exoskeleton use. Usability
studies run on non-workers may suffer from a bias, since they lack the perception and
acceptance assessment of the intended user. Introduction in the work environment brings
in further constraints in the exoskeleton architecture and devices, nature of coupling to
the human and user’s acceptability.
Following a literature search and benchmarking of available commercial devices,
FCA has planned a testing campaign to evaluate the applicability, usability and imple‐
mentation of a passive upper-limb exoskeleton in working tasks. In total, 42 workers
from FCA plants volunteered to participate in the research study. The paper illustrates
the main aspects of the testing campaign. Potential issues associated to the implemen‐
tation of this device in the automotive industry and to the biomechanical load estimation
in exoskeleton-assisted work tasks are briefly addressed.
2 Experimental Protocol
2.1 Passive Exoskeleton
In the study, the Levitate exoskeleton [20] was used as presented in Fig. 1. This passive
upper-limb exoskeleton aims at supporting the arms of workers exposed to repetitive
arm motion and/or static elevation of the arms. The exoskeleton consists of a metallic
frame for the core and two armrests for the upper arms (elbow and fore harm are not
interested) and can be worn like a backpack. Mechanical passive elements along the
arms partially support the upper limb muscles and shoulder joints. The support progres‐
sively increases when raising the arm. The exoskeleton can be custom-fit by regulating
length of the core metallic frame, size of the armrests and shoulder and waist straps, to
ensure comfort and optimal performance.
Analysis of Exoskeleton Introduction in Industrial Reality 237
Fig. 1. The Levitate passive upper-limb exoskeleton (images available at: http://www.
levitatetech.com; accessed 02/03/2017).
2.2 Participants and Procedure
The experimental campaign was conducted at the ergonomics laboratory (ErgoLab) of
FCA Manufacturing Engineering and included static and dynamic tests to be performed
without and with the exoskeleton. In total, 42 healthy male FCA operators volunteered
for the research study. Workers were identified based on their anthropometry in order
to fit the specification of the exoskeleton prototype. The tester we used was a medium
size, corresponding to P50 American male, classical percentile approach. Other inclu‐
sion factors include no limitation in strength or musculoskeletal disorders at the upper
limbs. Participants were informed in full detail about the aim and nature of the study
and they signed an informed consent. The measurements were carried out in accordance
with the Declaration of Helsinki.
Upon arrival at the lab, the worker was introduced to the working principles of the
exoskeleton. Personal data and anthropometric measurements were collected to fill in
the database and to regulate the exoskeleton as for lengths and choice of the mechanical
passive element characteristics. Participants performed the tests, without and with the
exoskeleton, in the same day and were video recorded using a frontal and lateral camera.
Between tests, they were given an adequate time to rest and were allowed to familiarize
with the device before they were asked to repeat the tests with the exoskeleton.
At the end of each task, both without and with the exoskeleton, a cognitive ergono‐
mist held a semi-structured interview with the worker, aimed at understanding the
quality of the interaction with the device. Other tools included the Borg Scale [21], to
quantify the intensity level of the activities as perceived by the subjects when tasks were
performed without and with the exoskeletons, a usability metrics questionnaire and a
TAM 2 questionnaire [22], to analyse the technology acceptance in relation to perceived
ease of use and perceived usefulness. In a final focus group, the moderator involved
workers in discussing on the use of exoskeleton, focusing on positive and negative
aspects in relation to their work context, and on the desirable characteristics of the
device.
238 S. Spada et al.
2.3 Tasks
The experimental campaign included two phases. The first phase saw the participation
of 31 male FCA operators. Participant mean height and mass were 174.9 cm
(SD ± 2.3 cm) and 81.6 kg (SD ± 9.1 kg), respectively. Mean age was 51.5 y
(SD ± 4.7 years). Two workers were discharged because their anthropometric measures
did not fully comply with the exoskeleton size. Full description of this first phase of the
testing campaign is reported in [23]. The three tasks were designed for this first phase:
1. A static task, conceived as an endurance test to evaluate the potential benefit of the
exoskeleton on the onset of muscular fatigue during a static action. Workers were
required to stand upright with extended arms (90° with respect the trunk) while
holding a 3.5 kg car spoiler, placed on the forearm so to exclude the wrist. The end
of task was set at subject’s will or because of a significant change in posture.
2. A repeated manual material handling task, developed from the FIT- HaNSA “Waist
up” Protocol [24] to evaluate the potential benefit (lessen muscle strain) of the
exoskeleton during a manual material handling activity vs. possible restriction to
movements. While standing, subjects had to lift up/back down a 3.4 kg mass from
waist level to shoulder level, following the beat of a metronome (30 actions/min).
The end of task was set at 600 s. However, subjects could stop before time if expe‐
riencing fatigue, discomfort or if going off cadence three times.
3. A precision task, resembling a sealing operation, developed to evaluate the potential
benefit of the exoskeleton (lessen muscle strain, higher comfort rating and dexterity)
during a precision task involving a significant static load on the shoulder joint. The
subject was standing, with the predominant arm almost extended, and used a felt-tip
pen to trace a continuous wavy line between two pre-market traces on a paper fixed
on a billboard. Subjects had to complete five different rows, containing 27 arches
each, from shoulder height to an overhead position, without removing the felt-tip
pen from the billboard. They could stop before the end of the billboard if experi‐
encing fatigue or discomfort.
In the second phase of the testing campaign, 11 FCA team leaders were selected
to run additional tests. Participant mean height and mass were 177.2 cm
(SD ± 5.0 cm) and 81.1 kg (SD ± 7.3 kg), respectively. Mean age was 45.8 y
(SD ± 6.9 years).
Team leaders possess a wider knowledge of the different work activities
performed by different workers and are usually involved in the work methods and
work organization. Subjects were initially asked to perform the static task and the
precision task developed in the first phase and described above. These additional
tests ensure a larger data set and allow confirming what observed on the first 31
workers.
The team leaders were then asked to carry out extra tests, conceived to simulate
real-work tasks, for which the subjects are adequately trained. The tests focused on
the following operations:
4. Mounting the clips of brake hoses underbody. While standing with arms above head,
subjects had to insert and remove 14 clips underbody (Fig. 2).
Analysis of Exoskeleton Introduction in Industrial Reality 239
Fig. 2. Mounting the clips of brake hoses underbody without and with exoskeleton
5. Sealing underbody using the sealing gun. While standing with arms above head,
subjects had to use the sealing gun to complete a 10-m close loop for as long as he
can endure (Fig. 3).
These first two tests mainly addressed prolonged awkward arm postures that could
emerge from underbody work. The posture is cumbersome and subjects knew they
could stop any moment if experiencing any fatigue or discomfort.
Fig. 3. Sealing underbody using the sealing gun without and with the exoskeleton
240 S. Spada et al.
6. Mounting the seal on the rear door. While standing, subjects had to assemble and
disassemble the seal on the rear door (no roll forming) using both arms. The complete
task was to be repeated twice. The altimetry of the car frame was chosen so that
subjects carried out the task between shoulder and knee height (Fig. 4).
Fig. 4. Mounting the seal on the rear door without and with the exoskeleton
The mounting task was conceived to investigate the effective activation and de-
activation of the exoskeleton through repetitive movements that involve a wide range
of motion of the arms as well work to be performed with the hands at or below waist
level, beneath the area of intervention of the exoskeleton.
3 Results and Discussion
Quantitative and qualitative parameters were analysed according to the main aim of the
test. In endurance tests, posture maintenance was monitored during the trial as well as
by visually inspecting the recorded video images. A comparison of the postures without
and with the exoskeleton was assessed, with particular attention to the arms and the
spine. No substantial differences were found.
Table 1 reports mean and standard deviation values of the endurance time registered
for the static task for the 11 team leaders, without and with exoskeleton. The time varia‐
tion interval (Δt = T
EXO
− T
NO_EXO
) and the relative variation (Δt% = (T
EXO
− T
NO_EXO
)/
T
NO_EXO
) are also reported. The time variation Δt is very similar to the 62.3 s registered
for the 31 plant workers [23], while the relative variation Δt% outgrows the 31.1%
registered for the 31 plant workers due to a lower T
NO_EXO
.
Table 1. Results of the static task
TNO_EXO [s] TEXO [s] Δt [s] Δt%
Mean 156.5 222.9 66.5 52.5%
SD ±86.1 ±110.0
Analysis of Exoskeleton Introduction in Industrial Reality 241
Table 2 reports the number of arches traced by the team leaders without and with
the exoskeleton, as for mean and standard deviation. The increase in the number of traced
arches with the exoskeleton is rather significant, even though there is a ceiling effect,
since many participants were able to complete the 135 arches when wearing the device.
When the percentage increment is computed only on those participants who did not
complete the arches, it increases to 34.0%, in line with the percentage increment of 33.6%
observed for the 31 plant workers [23]. The level of precision increased with fewer
portions of the wavy lines falling outside the pre-marked traces.
Table 2. Results of the precision task
No arches
NO EXO
No arches
EXO
Δ No arches Δ% No arches
Mean 108.3 127.3 19.0 17.5%
SD ±31.7 ±18.1
The positive outcomes on increased endurance time and precision level were
confirmed for mounting the clips and sealing underbody. Workers positively judged the
exoskeleton and declared it helped them to carry out the tasks with less physical and
mental effort, although they recognized the posture was still awkward and luckily not
common in real work tasks.
In the mounting task, operators complain potential interference of the exoskeleton
with the car frame. Also, the coupling between the worker’s arm and the device was not
always effective in the wide range of movements of the arm requested in the task. Some
participants reported some difficulties when working with the hands at or below waist
level, due to the force they had to exert to maintain the posture against the device. As
expected, introduction in the work environment brings in further constraints in the
exoskeleton architecture and devices, nature of coupling to the human and user’s accept‐
ability (Fig. 5).
Fig. 5. Potential interference of the exoskeleton with the car frame
In all, workers judged positively the exoskeleton and declared it helped them to carry
out the tasks with less physical and mental effort. The device was perceived particularly
242 S. Spada et al.
useful in tasks requiring raised-arm postures and when precision was involved. In the
“reaction adjectives” questionnaire, workers chose positive adjectives to describe qual‐
ities like easy to use, innovative, easy to understand, effective, efficient, and useful. In
the usability metrics questionnaire, the positive characteristics had high score (>4)
implying a good human-device interaction, but, on the other hand, operators considered
the work-device interaction critical.
TAM2 results showed that workers assigned high values (>4) to items that refer to
perceived ease of use and voluntariness, while they assigned low values (<4) to items
connected with intention to use, image and job relevance. Focus group results were
similar: workers affirmed that the exoskeleton can be useful in carry out work activities,
but the use should be on a voluntary-base.
Before adoption in real work environments, more information on worker’s accept‐
ance of the devices and long-term use is needed. Industry experience can reveal obstacles
to worker’s acceptance that are not evident in a controlled laboratory environment. Also,
the introduction of such devices requires a deeper understanding of the biomechanical
workload in exoskeleton-assisted work tasks and of how to appropriately evaluate the
impact of exoskeletons on safe physical works. Holistic risk assessment methods, like
EAWS [25], are required to specifically address the risk factors influenced by the use
of passive exoskeletons like posture, in this specific case the posture of the shoulder,
request of force and static muscle effort.
4Conclusions
In summary, the present study demonstrated that passive upper-limb exoskeletons may
assist workers in activities that involve prolonged raised-arms working postures. In total,
42 workers from FCA plants participated in the research study. It was found that, by
wearing the exoskeleton, participants increased the endurance time and the level of
precision (when applicable) in the task execution. Positive feedbacks also emerged from
the workers’ interview, who chose positive adjectives to describe qualities for the
exoskeleton such as easy to use, innovative, easy to understand, effective, efficient and
useful. Still, during the focus group, workers affirmed that the use of the exoskeleton
should be on a voluntary-base.
The positive outcome of this first experimental campaign opens to interesting next
steps but also to questions concerning the potential impacts of these devices in the work
environment. Further research should be conducted concerning the issue of the biome‐
chanical load estimation in exoskeleton-assisted work tasks and how exoskeletons may
affect the risk assessment and work methods.
Acknowledgements. The authors wish to acknowledge Chiara Carnazzo, Valentina Gabola and
Marco Bechis for their valuable support during the experimental activities and the data analysis.
Analysis of Exoskeleton Introduction in Industrial Reality 243
References
1. de Looze, M.P., Bosch, T., Krause, F., Stadler, K.S., O’Sullivan, L.W.: Ergonomics 59, 671–
681 (2016)
2. Rosen, J., Brand, M., Fuchs, M.B., Arcan, M.: IEEE Trans. Syst. Man Cybern. Part A Syst.
Hum. 31, 210–222 (2001)
3. Rocon, E., Belda-Lois, J.M., Ruiz, A.F., Manto, M., Moreno, J.C., Pons, J.L., Trans, I.E.E.E.:
Neural Syst. Rehabil. Eng. 15, 78–367 (2007)
4. Zoss, A.B., Kazerooni, H., Chu, A.: IEEE/ASME Trans. Mechatron. 11, 128–138 (2006)
5. Yang, C.-J., Zhang, J.-F., Chen, Y., Dong, Y.-M., Zhang, Y.: Proc. Inst. Mech. Eng. Part C
J. Mech. Eng. Sci. 222, 1599–1612 (2008)
6. Jarrassé, N., Proietti, T., Crocher, V., Robertson, J., Sahbani, A., Morel, G., Roby-Brami, A.:
Front. Hum. Neurosci. 8, 947 (2014)
7. Belforte, G., Sorli, M., Gastaldi, L.: International Conference on Simulations Biomedical.
Proceedings, BIOMED, Computational Mechanics Publications, pp. 199–208 (1997)
8. Gopura, R.A.R.C., Kiguchi, K., Bandara, D.S.V.: A brief review on upper extremity robotic
exoskeleton systems. In: 2011 6th International Conference on Industrial and Information
Systems, pp. 346–351 (2011)
9. de Looze, M.P., Bosch, T., Krause, F., Stadler, K.S., O’Sullivan, L.W.: Ergonomics 139, 1–
11 (2015)
10. Abdoli-E, M., Agnew, M.J., Stevenson, J.M.: Clin. Biomech. 21, 456–465 (2006)
11. Lotz, C.A., Agnew, M.J., Godwin, A.A., Stevenson, J.M.: J. Electromyogr. Kinesiol. 19, 331–
340 (2009)
12. Graham, R.B., Agnew, M.J., Stevenson, J.M.: Appl. Ergon. 40, 936–942 (2009)
13. http://en.laevo.nl/. Accessed 3 Mar 2017
14. http://www.backquality.com/. Accessed 1 Mar 2017
15. Barrett, A.L., Fathallah, F.A.: Evaluation of four weight transfer devices for reducing loads
on the lower back during agricultural stoop labor. In: Annual International Meeting of the
American Society of Agricultural Engineers (ASAE), pp. 1–8 (2001)
16. Ulrey, B.L., Fathallah, F.A.: J. Electromyogr. Kinesiol. 23, 206–215 (2013)
17. Whitfield, B.H., Costigan, P.A., Stevenson, J.M., Smallman, C.L.: Int. J. Ind. Ergon. 44, 39–
44 (2014)
18. Abdoli-Eramaki, M., Stevenson, J.M., Reid, S.A., Bryant, T.J.: J. Biomech. 40, 1694–1700
(2007)
19. Bosch, T., van Eck, J., Knitel, K., de Looze, M.: Appl. Ergon. 54, 212–217 (2016)
20. http://www.levitatetech.com/. Accessed 1 Mar 2017
21. Borg, G.: Borg’s Perceived Exertion and Pain Scales. Human Kinetics, Champaign (1998)
22. Venkatesh, V., Davis, F.D.: Manage. Sci. 46, 186–204 (2000)
23. Spada, S., Ghibaudo, L., Gilotta, S., Gastaldi, L., Cavatorta, M.P.: Investigation into the
applicability of a passive upper-limb exoskeleton in automotive industry. In: 27th
International Conference on Flexible Automation and Intelligent Manufacturing, FAIM2017,
27–30 June 2017, Modena, Italy. To be published in Procedia Manufacturing (2017)
24. Macdermid, J.C., Ghobrial, M., Badra Quirion, K., St-Amour, M., Tsui, T., Humphreys, D.,
Mccluskie, J., Shewayhat, E., Galea, V.: BMC Musculoskelet. Disord. 8, 1–10 (2007)
25. Schaub, K., Caragnano, G., Bitzke, B., Bruder, R.: Theor. Issues Ergon. Sci. 14, 616–634
(2013)
244 S. Spada et al.