5.3.1.3. Working sequence. Insuffiffifficiently planned construction tasks
can be more hazardous to execute, especially if the work involved is
of an unusual nature[75], especially an inappropriate working
sequence. This may be due to inadequate method statements, poor
design of temporary work, layout plans, schedules or site investigation
[76]. The training of construction work sequence typically consists of
work and non-workpiece-related activities. The information retrieval
effffort often increases when using an action sequence manual in
complex and intricate processes. It could lead to mental tiredness and
proneness to commit errors and safety issues
digitalized information into the real workspace using AR can provide
workers with intuitive tools to implement correct work procedures,
especially assembly procedure with improved accuracy and safety as
well as error reduction. Based on the research done by Hou et al.[21],
trainees are prone to making mistakes when they relied on their
Automation in Construction 86 (2018) 150–162
157memory and manuals to complete assembly tasks. When the developed
AR tool was used, the learning performance of trainees signifificantly
improved, and fewer errors and safety problem were made. Gender also
could be an additional factor that can inflfluence the safety post-training
performance of novice assemblers[77].
5.3.1.4. High-risk equipment operation. Many accidents are caused by
the inappropriate operations of heavy construction equipment such as
construction cranes. They are commonly used on construction sites
while elegant and safe operations of the equipment are diffiffifficult for
trainees to achieve given the limited information and sensational
control features provided through current training aids. The non
immersive virtual operation is regarded as a low-cost and easy-to-use
operating system for equipment training[35]. Li et al.[61]and Guo
et al.[78]developed multi-user virtual safety training systems (MVSTS)
based on non-immersive game technology, which could simulate the
detailed process of dismantling and tower crane operations. They
further provide dynamic databases for workers to compare their input
with predefifined knowledge and rules. Given the development
mentioned above, however, the integration of computer simulator
with immersive VR environment could give a realistic simulation for
depth perceptual feeling. It is expected to reduce experiential
difffferences between virtual simulations and real operations. Juang
et al.[79]adopted kinesthetic and stereoscopic (KS) vision into a
virtual crane simulator (called SimCrane3D+) to increase training
effffectiveness and operation confifidence. A unique and more intelligent
study conducted by Rezazadeh et al.[80]comes out with a virtual
crane training system which can be controlled by using control
commands extracted from facial gestures to lift uploads or materials
in the virtual construction sites for disabled workers. Multi-users and
multi-views features were also considered by operation training studies.
Kim et al.[38]developed an AR system with collaborative and
interactive scenarios for construction equipment operation, which
could share the idea with other users in distant locations. A multi
users, multi-views and the collaborative working system was well
developed by adopting both VR and AR technology[81]. It provided
a more realistic and easy-to-use environment, including better
coordination between equipment and materials, a more intuitive
control interface, and integrations with structural analysis tools.
5.3.2. Challenges faced by VR/AR safety enhancement mechanisms
There are several challenges in developing appropriate VR/AR
safety enhancement mechanisms: (1) some VR/AR systems provide an
unsafe collaborative environment that could be used for multi-users to
do the same work. However, it could not involve multi-role (i.e., safety
manager, workers, operators) into this environment[61]to achieve a
complete project-level human-computer interaction interface. (2) There
is still a lack of further studies on more diversity unsafe/hazardous
behavior training by using VR/AR technologies, such as electric shock
and object strike are envisaged to be conducted,which could involve
more force sensors to make workers experience the force feedback. (3)
Learning and memory curve losses on safety information are sub
stantially reduced during a nonstandard work sequence. However, the
real-time work package could give more detail task-based information
(according to Work Breakdown Structure (WBS)) to on-site workers
[82]. In addition, combing construction work package with VR/AR
technologies may provide a seamless and integrated safety guide in an
unsafe working sequence. (4) The limitations of generating kinesthetic
vision and the dizziness effffect caused by the stereoscopic glasses always
affffect the actual performance in the process of high-risk equipment
operation. The following research concerns are likely to be addressed in
•How to design a multi-user environment or collaborative virtual/
augmented equipment platform for difffferent trades in one dynamics
Could some work modes (i.e., work package) be involved into VR/
AR to make the work sequence more foolproof style and make
workers' behavior more skilled?
5.4. Classifification of safety evaluation methods and major challenges
5.4.1. State-of-the-art studies of evaluation methods in VR/VR applications
The experiment is the most popular validation method in VR/AR-CS
related studies[48,72–74,77,79]. Additionally, thefifield-based and case
study validation process have also been conducted in VR/AR-CSfifield
[35,38,81]. During the implementation process of these validation
methods, the subjective (i.e., interviews and questionnaire) and objec
tive (i.e., performance time, the number of errors) evaluation methods
are always adopted[83]. From the perspectives of evaluation purposes,
the effffectiveness, usability, applicability and the level of sense of pre
sence could be regarded as most concerned evaluation effffect[15]. Ef
fectiveness was assessed by conducting a comparative experiment with
a VR/AR system based group and a traditional process based group,
concerning how effiffifficiently a VR/AR system can solve a particular safety
issue. Usability and applicability were evaluated by a series of case
studies. The former concerns whether the VR/AR system could be used
in solving this safety problem, and the latter concerns which subject
group (novice or experienced workers) or safety issues could be suitable
for VR/AR systems application. For all VR/AR related applications, the
level of sense of presence is always important showing the abilities of
engaging users and eliminating the sense of deviations between sce
narios of VR/AR systems and reality. It can be evaluated in system trials
of users through post-training questionnaire survey in terms of users'
satisfactions, loading on operations (such as the uses of NASA TLX
form)[84], or observations during the trials[85]. A clear taxonomy
could be summarized to generalize a hybrid evaluation method for
future VR-CS studies as shown inFig.11and a detailed application of
evaluation methods in construction safety will be discussed in the fol
5.4.1.1. Hazard identifification. VR technology acts as an experience
oriented method of safety hazards cognitive for difffferent construction
participants in various project/activity/work-task levels. The multiple
baseline testing was adopted by Albert et al.[42]which selected six
baselines based on occupations (mechanical, civil, maintenance,
electrical, structural, and insulation) to conduct the experiment. It is
a method for validating the effffectiveness of workers in hazard
recognition. However, workers in various trades face difffferent types
of hazards. Thus, not all the hazards could be presented in this
experiment, which results in an inaccurate rate to hazard
identifification. The empirical records analysis is a subjective
evaluation method, which could be useful for a specifific project to
identify the hazards by designers and builders together[24]. This is a
new vision to involve the non-construction related participants (i.e.,
designers) to evaluate the VR in risk recognition and identifification. It
could be an effiffifficient way to reduce hidden hazards in the design phase.
转载自journal homepage:www.elsevier.com/locate/autcon
