5.2.2. Challenges faced by VR/AR application domains
5.2.2.1. Hazard identifification. Current studies on construction hazard
recognition by the applications of VR/AR are still encountered some
challenges, including ineffffective experiment results, capability
difffference on risks assessment and hazard forecasting: (1) Hawthorne
Effffect (subjects improve behaviors under observed)[57]and Practice
Effffect (subjects improve behaviors under training) may have signifificant
adverse impacts on the experiments conducted by traditional VR-CS
systems. Hence, a real-time hazard database or dynamic VR
environment may avoid these issues[36]. Adding the fourth
dimension, time, to create four-dimensional (4D) environments[5],
along with representative audio effffects, may signifificantly improve
dynamics in simulating construction projects; (2) The subjects may
assess risk levels with difffferent weights given to the judgments of
accident severity[41]. An adventurist may underestimate specifific risks
Fig. 10.A Taxonomy of Using VR/AR Systems in Training and Education.
Automation in Construction 86 (2018) 150–162
156in construction, and a milquetoast may identify more hazards than
ordinary workers[58]. Besides, safety culture including legal liability
and safety regulations in difffferent countries/regions affffects worker's
attitudes to construction hazards. Therefore, it is necessary to conduct
comparative experiments among the various subjects, regions or
The following research questions are suggested based on this dis
•Which technologies could be involved in VR/AR to improve its real
time and human-interactive ability in hazard recognition?
•How to conduct a cross-regional experiment to evaluate di
safety cultures in using VR/AR systems for the risk identifification?
5.2.2.2. Safety training and education. VR training initially was used to
rehearse the construction process and learn the safety hazards in a risk
free virtual environment. Thereby, the information emerged in a virtual
environment could be understood and translated easily to workers.
However, some challenges are still existed, including limited hands-on
experience, low learning or memory curving, high cost in high-risk
work training: (1) Most of the current VR-CS are still the offff-the-job
training in the offff-the-site location[31]. These offff-the-job VR systems
provide workers with limited hands-on experience with real working
conditions, thus resulting in a quickly forgotten or ineffiffifficient
performance when they come to the job site[60]. A close-to-reality
and multi-scenarios 3D dynamics environment with time sequence,
location, responsibility and knowledge database are needed to do the
on-the-job training with multi-users, especially the fast and skillful
decision-making training[61]; (2) AR-based on-the-job systems are
suitable for safety information-intensive training tasks, which could
convert safety information directly from paper-based plans to actual
work[62]. However, some physical based training work (i.e., electricity
installation) may pose potential safety hazards either in short-range or
in long-range. The more teleoperated AR training systems could assist
workers tofifinish the physical-based dangerous jobs in a remote place in
The following research concerns are likely to be addressed in the
•Could the hybrid training methods that combine on-the-job training with o
ffff-the-job training be integrated into VR/AR systems to train
multiple trainees in the future?
•What kind of educational methods, theories, and tools could be
smoothly embedded into VR/AR systems to improve the perfor
mance of training and education?
5.2.2.3. Safety inspection and instruction. Challenges for safety
inspection and instruction contain low interoperability of VR/AR-CS
information and unskillful visual literacy of workers: (1) Lacking
standardization of ICT tools for safety instruction is recognized as the
barriers for more accurate and real-time safety inspection or instruction
[64]. A mismatch between the level of details (LODs) for BIM and AR
always exists, which would cause the rich information-based BIM
models not to be well-displayed in AR interfaces; (2) the visual
literacy skills should also be improved in order to have higher
performance while dealing with visualized objects. A preferable
guideline can be chosen as Visual Information-Seeking Mantra[65]:
overviewfifirst, zoom andfifilter, then details on demand. Therefore, the
following research concern is likely to be addressed in the future:
How to give a clear taxonomy for combining VR/AR with other ICT
tools on display and information retrieval of construction safety?
5.3. Classifification of safety enhancement mechanisms and major challenges
The comprehensive investigations about the causes of construction
accidents were investigated. It found that the major contributors that
cause construction accidents include hazardous site environment, un
safe workers' behavior, unsafe working sequence and high-risk equip
ment operation[51]. Related safety enhancement mechanisms against
these contributors through the uses of VR/AR systems are discussed
5.3.1. State-of-the-art studies of VR/AR safety enhancement mechanisms
5.3.1.1. Working environment. The hazardous working environment is a
workplace with abnormal hazards violating the prevailing safety
standards and considering unsuitable for work. Inadequate security,
broken working platforms and other means of accessing the workplace
are also included[8,9]. The VR-based safety training and rehearsal
program can offffer close-to-reality simulations for the hazardous
working environment[66,67]. The users can effiffifficiently rehearse
tasks, plan, evaluate and validate the construction safety operations
or immerse with difffferent kinds of hazards to ultimately promote their
abilities for hazards cognition and intervention[6,68]. It also
contributes to raising the situational awareness of workers,
equipment operators, and decision-makers on a construction project
even in a remote location[33]. Teizer et al.[46]adopted remote data
sensing and visualization technology to train workers through the tasks
of identifying safety issues. The learning performance with the uses of
the unsafe virtual environment was enhanced through the ease of
recording and visualizing the nearby hazards and assessing the learning
effffect[67]. A virtual safety assessment system (VASA) was successfully
developed and evaluated by trials and post-use interviews[69]. The
results indicated that VSAS contributed to pinpoint the weaknesses of
construction workers who have passed the traditional assessment
process of identifying safety issues in hazardous activities including
stone cladding work, ironwork as wells as cast-in-situ concrete work
5.3.1.2. Worker behavior. The lack of proper training is one of the
contributory factors to risky worker behavior. Workers who are not well
trained tend to be less capable of recognizing hazardous activities, even
if well-trained workers may have a negative attitude towards safety.
Loss of balance was identifified as one of the triggering body behavior in
fall-from-height incidents during construction work[71]. Most of the
unsafe behavior training studies focus on balance-control training in
order to reduce the risk of falls at elevation. Walk training on real
construction planks in an immersive VR system, Surround-Screen
Virtual Reality (SSVR), was developed by Hsiao et al.
analyzed the working environment and personal protective equipment
(PPE) to provide appropriate training constraints of workers' behavior
in the system. For example, shoe design can signifificantly affffect workers'
lateral stability during walking on narrow and tilted planks at specifific
elevation[73]. To reduce the possibility of losing balance, mechanical
vibration should be minimized when performing construction tasks at
height[74]
转载自journal homepage:www.elsevier.com/locate/autcon
