Digital skin of the construction site: Smart sensor technologies towards the future smart construction site

AR-based applications

The effectiveness of onsite information systems such as BIM can be extended with real-time communication and real-time information. AR is defined as a combination of real and computer-generated scenes, where the “location” context parameter (set by real-world location) determines the content of the computer-generated scene. AR is also a technology that can bridge the gap between the digital and the real world (Meža, 2014). The recent special issue (2013) on “AR in Engineering and Construction” (AEC) in the journal Automation in Construction shows the need for, and interest in, research and innovation in the field. In another relevant study, Ibrahim and Kaka (2008) presented a review of photographic/imaging applications in construction. In addition, a 2013 survey (Chi et al., 2013) identified future trends of AR in AEC applications, including field exploration through localisation, accessing field information using ubiquitous services and integrating with location-specific information. With the advancement and widespread of BIM in the construction industry, AR applications are becoming increasingly useful (Wang et al., 2014; Park et al., 2013; Meža et al., 2014). The prototype developed by Wang et al. (2014) demonstrates the ability to use BIM + AR for visualisation/walking through, information retrieval, onsite assembly and way-finding based on the “location” context parameter. In a study by Wang et al. (2014), an AR prototype was developed to assist the installation of piping. Similarly, Meža et al. (2014) reviewed system functionality from a technical point of view. The prototype by Meža et al. (2014) was tested on the construction site of a 12-storey residential building. Park et al. (2013) proposed a framework for construction defect management using BIM and AR based on ontology. Although their study was limited to the conceptual, laboratory level, it proposed an innovative, real-time, location-based, proactive defect management mechanism.

Industry research and/or products have also come out recently which deliver contract-awareness for information visualisation. An AR smart helmet (DAQRI, 2017) has been developed for next-generation construction and is currently undergoing industry trials. The smart helmet, when integrated with sensor and imaging technology, can create a location-specific AR overview. Functions of the helmet include situational awareness through data visualisation, thermal vision, guided work instructions and remote expert assistance. The developers of the prototype helmet, which has already been tested in a number of sectors, are currently seeking trial opportunities in the construction industry. Furthermore, an AR-based mobile app has been developed (Dalux, 2017) and trialled in construction projects. Through the app, construction workers can view the physical environment in real time, allowing them to visualise and interact with it in various ways. Despite limited information on trials being publicly available (regarding whether it has been verified through developmental or operational evaluations and tests), the technology appears to be available on the market. Trimble (2017), in collaboration with Microsoft, is developing mixed-reality applications for the building and construction industry. Pilots of such applications have been conducted with AECOM (CIOB, 2016) on three construction projects in London, Hong Kong and Denver. Also, Kamat and Dong (2017) invented a visualisation method which blends real and virtual construction jobsite objects in a dynamic AR scene in real time. The steps of their process include capturing an image of a construction jobsite object through a depth-sensing device and colour camera, registering the location of those images in a common coordinate system, projecting geographic information system (GIS) and/or CAD data to generate the AR scene, and removing hidden surfaces. Evidence on applied investigation of this technology is yet to be reported.

No matter how fascinating the technology is, for it to be used to its full capacity it must be accepted by the end users. In order to clarify the social aspects of AR in n-dimensions (Edirisinghe and Zaslavsky, 2014), a more generic representation of context parameters has been proposed by Kim (2013). Kim (2013) specified three dimensions of context immersion in AR: communication, relationship and mobility.

BIM-based applications

A conceptual framework linking BIM-based AR to construction sites was proposed by Wang et al. (2012). In their study, Motamedi and Hammad (2009) proposed a conceptual framework of radio-frequency identification (RFID) tagging linked with BIM for progress monitoring of construction projects as well as for facility management research.

BIM has been studied for the purpose of linking onsite information and activities in various aspects of construction, such as scheduling (Kim, Kim and Kim, 2013; Kim, Anderson, Lee and Hildreth, 2013; Wang et al., 2014), supply chain (Aram et al., 2013) and quality assurance (Chen and Luo, 2014). While real-time context was not used in these studies, location was manually tracked in some of them (Kim, Kim and Kim, 2013; Kim, Anderson, Lee and Hildreth, 2013; Aram et al., 2013). An industry experiment conducted with augmented drones to facility construction progress monitoring (Bentley Systems, 2016) was not live. Similarly, there is little evidence about real-time capability of the AI technology that spots health and safety breaches on site (CIOB, 2017d) even though it was trialled by a number of contractors. Because these studies were not intended to allow for real-time behaviour adaption, they are not relevant to the concept of context-aware digital skin and are therefore excluded from this review.

However, these studies include extensive useful onsite context parameters – information related to site activities, resources (labour, material, plant and equipment), durations, products and processes – giving them the potential to qualify as digital skin if they were to be extended with real-time context capture and adaption. For example, smart BIM (Heidari et al., 2014), which is a prototype of real-time interaction and task performance linked with smart objects and BIM, can be considered part of digital skin. Another example is the iHelmet (Yeh et al., 2012), which extends the safety helmet into a real-time information retrieval and projection device. An iPod and a projector are attached to the safety helmet of a worker in order to retrieve BIM information for the worker when their location is entered. This information is then projected onto a nearby surface. This work is also an example of the use of wearables on smart construction sites. Akinci et al. (2006) propose an innovative approach of using temperature-sensors-based 3D point cloud-based system for construction quality control. The detection algorithm was verified through as-designed and as-built model comparisons of four cases.

3D mobile mapping technology (GeoSLAM, 2017) based on simultaneous localisation and mapping with handheld scanners has been trialled in a complex renovation project in the USA. An accuracy of 2 cm was recorded, with a four and a half hour time to scan the case study site. A combination of hardware and a mobile app called Spike (IkeGPS, 2016) converts a smartphone into a laser rangefinder which is compatible with commercial BIM software. A laser rangefinder hardware device is clipped onto the smartphone and communicates with the phone using Bluetooth. This smartphone-based laser rangefinder is an industry product that allows real-time measurements to be captured on site. A BIM-aware location-based application has also been developed for mobile devices (Dharwada et al., 2016), which includes the ability to display floor plans based on the device’s location, representations of equipment, real-time status information and alerts about physical equipment based on the location of the equipment, as well as information from a number of onsite sensors for parameters such as air flow, temperature of air within the VAV box, temperature settings and humidity.

Labour tracking applications

Navon and Goldschmidt (2003) proposed a conceptual model for automated data collection technology to locate construction workers; this was later used by Sacks, Navon and Goldschmidt (2003) to test a prototype labour control model integrated with a building project model.

Khoury and Kamat (2009) evaluated three position-tracking technologies for user localisation in indoor construction environments: WLAN, UWB and indoor GPS positioning systems. The proposed indoor GPS system uses laser and infrared (IR) light. The precision of each system was evaluated and the technical criteria, logistical issues and costs were discussed. The authors concluded that decisions on technology should be based on important technical criteria such as calibration and line-of-sight, in addition to other logistical matters such as availability, the prevailing legal environment (e.g. permitted bandwidth) and the associated implementation cost. The testing in this study was conducted in a controlled environment.

Ubiquitous location tacking systems to deliver context-specific information have been proposed for construction sites using integrated WLAN and GPS as the base technologies, such as by Behzadan et al. (2008). The tracking application proposed by these authors can automatically switch between positioning technologies based on whether the user is indoors or outdoors. Despite the practical limitations imposed on construction workers by wearing bulky devices, the proposed project is a viable approach to delivering context-specific information based on the user’s location on the construction site. The capability of the system to deliver context data was verified in a controlled environment by Yang et al. (2010), who used a video camera machine learning algorithm to track workers on site. They developed an algorithm to track multiple workers and tested it in an experimental set-up as well as two real-world site scenarios, though neither of these real-world sites was highly dynamic.

More recently, labour tracking concepts have been extended to the monitoring of productivity (Cheng et al., 2013) and to ergonomic analysis of construction sites (Cheng et al., 2012), which has health and safety implications. Cheng et al. (2013) proposed a system that automatically analyses workers’ activities task-by-task. The system uses a combination of real-time location sensors and thoracic posture data to analyse worker’s activities by encoding material handling, and idle and travel in various zones, including the work zone, storage zone and rest zone. Despite the fact that the experiments were conducted in a controlled environment, rather than a real-world setting, the technology seems promising. Similarly, Jiang et al. (2015) developed a labour consumption measurement system based on GPS. Their system processes location data to determine whether a labourer is active within the boundaries of a predefined construction region. This system has been applied in a large hydropower project. To achieve the same goals in an interior construction setting, a vision-based method using special sensors would be applicable. However, this would be a particularly challenging task as construction activities have a large range of intra-class variability, including varying sequences of body posture and time spent on each individual activity (Khosrowpour et al., 2014).

Passive RFID tags were attached to the hard hats of construction workers to track when they entered or left a site (RFID Journal, 2011). Management was able to capture the identity of each worker through a tag linked to the worker’s name, address and other details, such as the worker’s employer and training history. This tracking solution appears to be an established industry product which is commercialised through a monthly service fee.

Cheng et al. (2012) used the real-time location sensors and thoracic posture data combination to analyse the ergonomics (Migliaccio et al., 2012; Ray and Teizer, 2012) of construction workers with a view to detecting unsafe behaviour in materials handling. A Wi-Fi finger-printing-based localisation technique was tested for indoor tracking by RFID tagging workers’ safety helmets (Woo et al., 2011). An innovative, non-invasive, device-free detection and localisation technique has also been proposed for construction sites (Edirisinghe, Blismas, Lingard, Dias and Wakefield, 2014). This study proposed the use of Wi-Fi signal processing techniques to mitigate the multi-path effect caused by the movements of construction workers onsite, especially detecting when workers had entered hazardous zones.

A construction hard hat with attached electronic circuitry tracks workers and their activities on site for safety objectives (Hudgens and McDermott, 2009). The electronics are detected by radio-frequency sensors placed throughout the construction site, as long as the hats are in range. Once detected, the location of the hard hat is determined and personal information stored in its electronic circuitry is received by the sensor. The location and personal information is reported to a monitoring system that uses it to track personnel movements, to detect unauthorised activity and to generate reports.

Supply chain applications

Research has identified the supply chain as a critical element of the digital skin, relevant both for productivity and security reasons. Work on automating the supply chain began with the introduction of barcodes in the construction industry for various purposes such as quantity take-off, field material control, warehouse inventory and maintenance, issuing of tools and consumable material, timekeeping and cost engineering, purchasing and accounting and document control (Bell and McCullouch, 1988). Bernold (1990) reported on the use of barcodes for material management in construction; an approach developed in a yard control system. Since then, substantial research has been conducted on the tracking of material and tools; the most widely used technology for this purpose is RFID. Li et al. (2005) proposed combining barcode and GPS technologies for material and equipment tracking in order to reduce construction waste. Jaselskis and El-Misalami (2003) and Song et al. (2006) used RFID for material tracking, and Goodrum et al. (2006) used the same technology to track tools. Ergen et al. (2007) used combined RFID and GPS to locating precast concrete components on site. Jang and Skibniewski (2009) proposed asset tracking based on combined radio and ultrasound technologies. Accuracy assessments were conducted on UWB technology to track construction resources in indoor (Maalek and Sadeghpour, 2013) and harsh environments (Cheng et al., 2011). Irizarry et al. (2013) integrated BIM with the GIS to enable tracking of supply chain status and to provide warnings to ensure the delivery of materials. Although the system was intended to automatically visualise the data in real time, data were entered manually in the case study. Recently, Montaser and Moselhi (2014) extended RFID technology to track workers and material non-invasively by analysing the signals using a signal processing technique called triangulation.

A mobile app tracks the material deliveries by calculating their volumes (CIOB, 2017c). A patented image processing technique (Boardman et al., 2016) is used to estimate the measurements without the need for any additional hardware, unlike other specialised measurement techniques that require special equipment. Other than this, no information is available about the trials of the app in construction, even though it is available as a commercial product. A solution based on RFID tags has been used to record prefabricated units in Hong Kong (RFID Journal, 2013a). In addition to the benefits of tracking the components as they leave the factory and are installed on site, being incorporated into a new structure, the method also reportedly avoids onsite errors, ensuring quality.

Mobile equipment operation applications

An early study on the control of construction plant using GPS is by Roberts et al. (1999). This work proposed using GPS to automate a blade laser control system in a bulldozer. Sacks, Navon, Shapira and Brodetsky (2003) developed a conceptual roadmap for automated monitoring of construction equipment using GPS data to monitor concrete bucket loading and unloading operations. This was later field tested in crane operation (Sacks et al., 2005). Oloufa et al. (2003) applied GPS-based tracking of vehicles on construction sites to avoid collision. Lu et al. (2007) proposed RFID-based tracking and positioning of construction vehicles by integrating GPS with a vehicle navigation technology called “dead reckoning” (DR). DR automatically supplanted the GPS when GPS signals were unavailable or unreliable.

Pradhananga and Teizer (2013) implemented a GPS-based tracking and analysis system for construction site operations. In addition to the speedy analysis of single-system equipment, their GPS-based approach allows proximity analysis (Ray and Teizer, 2012) of multiple sources, which is useful for determining risks on job-sites. The tracking applications of GPS would be particularly useful in situations where multiple mobile plants are in operation. In a study by Ray and Teizer (2013), the proximity of a skid steer and an excavator was tracked and analysed for an entire day. The authors later used laser scan data to generate a through-point cloud to test blind spot monitoring for construction equipment.

Lee et al. (2012) proposed a BIM-based tower crane navigation system for blind lifts. The system they developed shows the location of a lifted object in the context of a building and its surroundings using an imported BIM model and data collected through various sensors including a video camera. The experiments were conducted on the construction site of a research building on a university campus. The technology’s acceptance was also evaluated using two aspects of the technology acceptance model (TAM) by Davis (1989): ease of use and usefulness.

Reader-based RFID tags were affixed to tools, materials and components during the construction of an oil refinery in Australia (RFID Journal, 2012). Although information is not available as to the type of RFID tag used, the trials provided some useful results and user perspectives about acceptance of the technology. The suggestions included reducing the number of readers and obstructions to construction traffic, and that the system ought to be made robust enough for harsh site conditions and extreme temperatures. RFID tags were piloted for machinery and equipment tracking by the North American Construction Group (RFID Journal, 2009). Handheld readers were used to detect pieces of equipment valued over $5,000. RFID-tagged equipment on geographically dispersed sites were located using Google Earth software. An Italian engineering and construction company tracked offshore equipment such as vessels, cranes, drilling rigs, steel pipe slings, shackles and buoys around the world (RFID Journal, 2010). A handheld reader was used to locate the tags, on which important data about the equipment had been stored, including inspection dates and insurance information.

While a tremendous number of studies have been conducted on vision-based video or photography analysis and image processing techniques for progress monitoring and scheduling (Fard and Peña-Mora, 2007; Azimi et al., 2011), this paper focuses on sensor-based applications in line with the concept of digital skin. Costin et al. (2012) used RFID tags to track equipment, material and workers on a high-rise renovation project to provide activity and project status monitoring. Vision-based techniques were also applied to track equipment operation. Automated 2D detection of construction workers and equipment through onsite video streams was conducted by Memarzadeh et al. (2013). Tajeen and Zhu (2014) successfully applied computer vision techniques to distinguish five classes of construction equipment (excavators, loaders, dozers, rollers and backhoes).

Schedule and progress monitoring applications

Hendrickson and Rehak (1993) and Retik and Shapira (1999) were early researchers on the visualisation of site activities using real-time images. Retik and Shapira (1999) integrated site-related activities into the planning and scheduling of a construction project using virtual reality. Reinhardt et al. (2000) proposed a system architecture for construction progress monitoring. Cheng and Chen (2002) combined barcode and GIS technologies with video monitoring to develop an automated schedule monitoring system for precast building construction. Fard and Peña-Mora (2007) dealt with visualisation techniques for construction progress monitoring that are not automated. Their technique compared a model of the building “as-planned” with a photograph of the building “as-built”. Lee and Peña-Mora (2006) analysed progress through a material-based detection technique and the generation of an occlusion-free photograph. Chin et al. (2008) integrated RFID and four-dimensional computer-aided design to assess progress on structural steel works in two real-world high-rise construction projects. In their research, Rebolj et al. (2008) developed an automated activity tracking system based on image recognition. A pilot implementation of their system was used in an industrial construction process. Shape recognition was used as the image processing technology in construction by Brilakis and Soibelman (2008). Son and Kim (2010) field trialled a stereo-vision-based automated progress monitoring system to model the structural components of steel buildings. In research by Roh et al. (2011), progress monitoring was undertaken through the visualisation and comparison of as-built photographs of the construction project to an as-planned 3D BIM, together with a walk-through feature. Virtual (as-planned) and real-world (as-built) domains were integrated in this research; however, the photographs of the as-built structure were captured manually. The focus of Yang et al. (2012) was tower crane activity, which they tracked through image processing of video surveillance camera data. Their system was tested in two construction scenarios: concrete pour cycles and material hoist cycles.

Drones were used to capture video footage to analyse construction progress in real time (MIT, 2015). The technology was trialled to monitor the activities on a number of construction sites. User acceptance of the technology and issues around privacy were not reported. A multi-screen AR system invented by Hsieh et al. (2016) was used to monitor construction processes on site. It was equipped with multiple touch display screens to aid decision making by separating public and private information for the project, so as to improve efficiency and reduce misunderstandings during the construction process. A method for monitoring construction progress through image processing was invented by Golparvar-Fard et al. (2015). In this method, photographs taken on site are used to construct a 3D “as-built” model, and this is compared with the “as-planned” model. The as-built model is overlaid on the as-planned model for joint visualisation to show progress towards completion of the structure. Progress is indicated with colour-coded changed and unchanged elements in red and green, respectively, using the D4AR modelling platform (Golparvar-Fard et al., 2009).

Safety management applications

Every resource on a construction site introduces its own set of safety hazards and exposes safety risks. Advances in ICT allow for representation of and reasoning about construction context parameters for safety management, which is part of the concept of digital skin.

In Wu et al. (2010), a real-time tracking system was developed as a low-cost solution to track near-miss incidents. This system included ultrasonic transceivers for outdoor and indoor real-time location tracking, adopted sensors for environment surveillance and used RFID for access control as well as storage of safety information about workers, equipment and material and a wireless sensor network for data transmission. The sensor boards were equipped with light, noise, temperature and humidity sensors. Historical near-miss accidents were analysed based on typical accident cases, and features of construction sites and characteristics of near-miss accidents were also considered in the development of the system. Field trials were conducted in a multi-storey warehouse. The paper argued that a warehouse sufficiently simulates the environment on construction sites. However, it is arguable whether the dynamicity and practical limitations of a construction site, in terms of activities and processes, are similar to those of a warehouse, especially for the evaluation of near-miss incidents, even though they are structurally similar. Teizer et al. (2010) tested the same RFID technology on a clean-coal power plant construction site to track the proximity of construction workers and equipment operators as part of a safety alert system. A recent study by Teizer and Cheng (2015) used UWB and GPS sensor location data of workers, equipment and georeferenced hazard areas to automatically identify additional hazards and to analyse the spatial-temporal conflicts between workers and the identified hazards. Experiments were conducted in a laboratory environment as well as in the field on a live site.

With advances in wearable technologies in the wider pervasive environment, these devices are also emerging in the construction industry. Edirisinghe and Blismas (2015) propose an innovative approach using wearable technology and e-textiles to monitor the physiological parameters of construction workers in real-time to improve health and safety. The technological feasibility of the proposed system has been validated by a prototype system that monitors body temperature, detects heat-stress conditions of construction workers and generates early warnings. Similarly, Yi et al. (2016) propose using existing wearable technologies combined with an evaluation of perceived exertion to improve health and safety for workers in hot and humid environments. This system includes wearable wrist-bands for the workers and environmental sensors to generate warnings. In the experiments, GPS was used in the outdoor setting of a construction site.

A wearable kit was designed to improve the safety of construction workers by means of real-time tracking of context parameters, and this kit has passed the field trial phase (CIOB, 2017a). The collection of online equipment was designed to improve the safety of operatives in the field, as well as the ergonomics of working environments. The online operative kit/sleeve was designed to have multiple personal protective equipment (PPE) elements to suit the operational requirements of project partners. These elements include: a high-vis jacket equipped with sensors able to analyse the air; glasses with an integrated camera; and safety boots allowing geolocation.

The EskoVest is an upper body exoskeleton designed to improve the health and safety of construction workers (Eskobionics, 2017). This robotic vest enables construction workers to lift heavy objects while minimising their exposure to muscle strain and injury. This commercial product is ready to launch (CIOB, 2017b). More details about the wider adoption of the technology in the construction industry are yet to be reported, possibly due to the disruptive nature of the technology, which features lots of electronics and control systems and which may present challenges to do with worker training.

Kelm et al. (2013) used a novel application of RFID technology to improve safety. Low-cost RFID tags were attached to PPE which was then combined with personnel identification to check how well personnel were complying with PPE requirements and to provide users with timely feedback. Personnel entering a construction site through a site gate were informed of the safety process and procedures to follow.

SmartCap (2017) is a commercially available product designed for fatigue monitoring. The safety helmet has a headband that monitors brain activity to capture signs of fatigue. This identifies times of risk of microsleep – unintended episodes of loss of attention often caused by monotonous or repetitive tasks. Alerts are sent through a mobile app. Site testing has revealed a reduction in fatigue incidents, as well as technological, social and policy challenges around wider adoption, including training workforces with limited literacy, calibration issues due to staff turnover and policy requirements due to union presence.

A ubiquitous sensor network (USN) was prototyped to monitor safety during concrete placement in a study by Moon et al. (2015). The system combined multiple technologies, such as sensors, wireless networking, safety monitoring applications and an independent power supply. The sensors collect context parameters such as distance, strain, weight and inclination. The safety manager in the site office can oversee the safety of concrete formwork and be assured that the construction operation is being safely executed through a web interface, smart phone and/or a smart glasses application. An independent power supply is used for USN technology to address the energy challenge which is common in wireless devices. The technology was tested in an onsite case study of concrete column construction. An RFID-based solution was used to automate safety and access management on site (RFID Journal, 2013b). Workers were given photo ID cards with built-in passive RFID tags linked with their individual details. A zone-based site access system based on safety training was also implemented. The RFID-based apparatus was proposed for mobile equipment to ensure the safety of workers in proximity. The apparatus is composed of an article of clothing wearable by the workers (clothing, a vest or a hard-hat) fitted with an RFID tag with an antenna. A second sensor is mounted on the mobile machine to detect the proximity of the RFID tags (Dasilva and Shervey, 2012). The plurality of directional antennas each corresponding to a danger zone sends and receives signals to and from the RFID tags within each zone.

Visualisation-based tracking to applications to improve safety

Some research combined tracking with digital engineering or visualisation for the purpose of improving safety. While serving the purpose of safety management, these technologies are also linked with visualisation. Hence, they are categorised separately.

Early work in this area included a system implemented to automatically monitor fall hazards through guardrail analysis (Navon and Kolton, 2006). The automated model identified fall hazards during scheduled activities through sensors which monitored the existing protective measures in real-time and warnings were generated accordingly. The proposed technique was implemented in a prototype for one kind of protective measure (guardrails) and tested onsite in an ongoing project where use feedback was also tracked. In a later study, Lee et al. (2009) developed a monitoring system based on ultrasonic and IR sensors combined with a wireless telecommunication system to detect fall accidents. The effectiveness of this system was evaluated on a real construction site. Real-time location data collection and visualisation technology were applied in construction safety and activity monitoring by Cheng and Teizer (2013) using crane lifting as a case study.

Park and Kim (2013) developed a prototype of a safety management and visualisation system with BIM, location tracking and game technologies. A prototype system has been developed and tested, based on an illustrative accident scenario. Once the safety risks are registered by the system during inspections, they are visualised in real time. Safety education can be facilitated through a question-and-answer-based game. This system was evaluated in a case study. A prototype for safety education was implemented for pipe shaft opening. This prototype was evaluated by construction workers, safety managers, construction managers and trade leaders on three criteria: identifying risk before work execution, increasing comprehension of accident risk and real-time communication.

The CoSMoS system (Riaz et al., 2014) is a prototype developed to visualise the environmental temperature of confined spaces through real-time sensor data to improve safety. The prototype deploys two TelosB (Levis et al., 2005) sensor motes in two confined spaces. Sensor data are visualised in the BIM in real time and the system generates warnings accordingly. CoSMoS is an example of a context-aware pervasive system that can adapt to its environmental context. The study reported that the accuracy of sensor data is a major challenge. The prototype was used to monitor confined space using only two sensor nodes. Industry feedback was captured in a focus group with contractors, researchers and consultants. The system’s effectiveness, proactiveness, practicality, usability, financial feasibility, potential for future improvements, benefits and implementation barriers were evaluated.

Zhang et al. (2015) developed a prototype of automated workspace visualisation in BIM, using a remote sensing approach for activity-level construction site planning that can proactively improve construction safety. The system senses location data through a GPS attached to the hard hats of workers, and gives an approximation of the workspace to detect potential workspace conflicts among competing work crews or between material lifting equipment. Images from both a video camera and an unmanned aerial vehicle were also analysed. The prototype was trialled on a real site for three days during concrete column construction activities.

Carbonari et al. (2011) used UWB-based position tracking to implement virtual fencing to improve safety. The system puts in place a safety policy that generates warnings to prevent workers accessing hazardous areas (where the spatial coordinates of the area are predefined), such as where there is a high likelihood of objects falling from overhead. The prototype was tested under laboratory conditions as well as in the field.

Future construction site safety education is also part of digital skin. For example, in a study by Teizer et al. (2013), safety training for steel workers was conducted using real-time location tracking data visualisation technologies, putting the system to the test in the construction worker education and training environment.

Automatic safety control

Automatic safety management on the future construction site is illustrated in Figure 7. Site safety and access management will be automated (RFID Journal, 2013b). Site environment conditions such as temperature, humidity, ultra violet light and air pollution levels will be monitored and visualised on BIM (Riaz et al., 2014). If the conditions are unacceptable or unsafe, management and workers will be warned (e.g. of fire, as shown in the figure). Work exclusion zones will be monitored in real time and access to unauthorised areas will be tracked (Edirisinghe, Blismas, Lingard, Dias and Wakefield, 2014) (e.g. unsafe lifting operation, unsafe access/egress, as shown in the figure).

Workers will be alerted to possible falling objects as a result of unsafe lifting operation, as shown in Figure 7) from overhead (Carbonari et al., 2011). As discussed in the plant operation section, plant and equipment will carry tags so that overloaded vehicles, plant operating in close proximity to workers or multiple plants will be tracked, reducing risk of collision. Since almost every object on site will be part of the digital skin, unsafe working heights, open shafts and edges, unguarded machinery, unsafe scaffolds and unstable platforms will be tracked and risks to workers will be controlled.

Health and safety communication

Safety communication includes orientation training, short refresher training and the communication of rules, procedures, policies and legislative requirements for safe ways of performing day-to-day work procedures. Examples of such requirements include safe work method statements or job safety analyses for high-risk activities in Australia, or risk assessments for high-risk activities in Singapore. All these training materials will be part of the automated safety control system. All such health and safety policies, regulations and related documentation will be part of the automatic safety control system.

Video-based health and safety training has been identified as a more efficient way to communicate policies, procedures and rules to construction workers due to potential literacy and comprehension barriers (Lingard et al., 2015). On the future construction site, safety training and tool-box meetings will be more interactive (to suit generational and adult learning needs) and also more visual; training will be conducted using real-time data visualisation (Teizer et al., 2013) or by scanning a code to play a video (Lingard et al., 2015).

Limitations of hardware/sensors and software/applications

As introduced in Section 3, the key technological elements of the future smart construction site are the hardware, communication technologies and software. Amongst these, the essential and integral parts of the smart construction site are the sensors and hardware. However, the low reliability, low computational power and limited battery life of sensors and devices present challenges. The sensitivity of sensors and the accuracy of sensor data significantly affect the reliable functioning of any system. For example, the smart bracelet used by Yi et al. (2016) for monitoring heart rate suffered a transient malfunction during site testing. Similarly, the accuracy of sensors (Chen et al., 2013) and GPS due to their dependency on external systems has been reported by many researchers as a limiting factor (Meža et al., 2014; Jiang et al., 2015).

In addition, as Meža et al. (2014) reported, one of the limiting factors for using AR applications is the limited hardware capability of mobile devices. About a decade ago, Ahsan et al. (2007) reported that battery power was a technological limiting factor, among many others, to the use of ICT in construction. Wireless sensors are also battery powered and still, as of the time of writing, have practical limitations to do with onsite longevity. In addition, there are also limitations specific to particular pieces of software. For example, geo-referencing and visual occlusion are challenges for AR applications (Meža et al., 2014).

Thorough technology validation

Innovation is the key to cutting-edge research. However, rigorous validation techniques should be followed if the technology is to be used to its full potential. The TRL model recommends that technology be validated at the stages including: pre-concepts and basic ideas though proof of concept; applied investigation through laboratory or simulated environment; and development and operations through operational environment testing.

Technology feasibility testing should include thorough laboratory testing or experimental system evaluations in simulated environments (controlled set-ups) prior to taking the technology into a real-world setting. For example, in the study by Yi et al. (2016), the transient onsite malfunction could have been discovered during prior laboratory testing.

The next phase of technology validation in a real-world site environment should cover aspects such as ruggedness (Ahsan et al., 2007), seamless onsite integration, better understanding of the demands placed on the technology in practice and a wide range of practical issues, such as the directions and ranges of radio waves on real-world sites (Goodrum et al., 2006). For example, Meža et al. (2014) discovered practical onsite challenges which are common in reality when the AR technology was trialled that would have been difficult to anticipate in a simulation. Onsite testing can prove the robustness of the proposed system.

There are other factors to consider in technology validation, such as generalisability and scalability, which are covered in stage 7 of the TRL model, full, scaled-up system demonstration (pre-product with all requirements) in an appropriate generalisation of real or operational environment, which is part of the development and operations phase. Generalisability is the degree to which the system can work on any construction site. Once site tests validate the technology on a single site, it is important to consider whether the technology’s effectiveness depended on any site-specific characteristics. Technology should be fully generalisable and applicable on any site. Scalability refers to the ability of the system to grow. For example, the scalability of a sensor network system with a pair of nodes can be tested by increasing the number of nodes that it uses. This may introduce a range of other challenges, such as problems with communication protocols, timing for communication cycles, network reliability, packet loss and resilience, etc., which are important to address to create a robust system. For example, Riaz et al. (2014) conducted testing in a real environment, but included only two sensor nodes. Such systems should ideally be scaled up in testing to realise the adoption of technologies to their full potential.

User requirement analysis

It is vital to understand what the technology users’ expectations are before a product/system/technology is developed. In software engineering, for example, user requirement analysis is a major phase in the development cycle. Similar user consultation prior to onsite implementation is critical in the construction industry if the technology is to have the desired impact. None of the research studies reviewed has gone through this process. One reason could be that researchers from an industry background assume that they already know the industry’s expectations. Kang et al. (2013) propose information technology (IT) “best practice”, under which practitioners contribute to technology development prior to implementation by considering the processes and effects of the systems on project performance directly. Half a decade ago, Erdogan et al. (2010) argued strongly that “user-centred IT” should be developed for the construction industry. In their view, many of the promises to meet the needs and enhance the experience of ICT systems in the industry remained (and remain) unfulfilled. The present paper argues that a major part of the problem is that industry practitioners do not become involved early enough in system development so that the researcher can capture their requirements thoroughly, as would occur in software engineering.

If the technology is cutting-edge, requiring an intermediate RtP step between prototyping and real-world testing, user requirement analysis can happen then; user needs should be captured before the technology is taken to the site. The ultimate goal of this exercise is for the technology to make an impact in the industry. As Erdogan et al. (2010) argue, this is needed to increase the ability of organisations to adapt and improve upon current processes by adopting new technology.

User acceptance testing

After the technology is implemented and trialled in the industry, it is important to test whether the system meets the objectives and expectations of users. People transformation, managing change and fostering innovation, especially in the construction industry, which lags in technology adoption, can be very challenging. The social aspects of technology adoption in the construction industry have been studied by some researchers (Xu et al., 2014; Levis et al., 2005; Cao et al., 2015). Some studies reviewed attempts to validate technology adoption by the users (Lee et al., 2012; Jaselskis and Misalami, 2003). Jiang et al. (2015) note that full automation of the proposed technology is not possible due to unpredicted circumstances on site, where a resident supervisor’s knowledge is critical. In another study on BIM, Xu et al. (2014) argue that improvement in the mode of thinking of adopters as well as potential adopters is the most important driver for wider adoption of BIM in the construction industry. In a similar study on BIM and AR, Wang et al. (2014) suggested that contractors should be actively encouraged to embrace AR technologies’ intended uses and capabilities or they could find themselves entrenched in their comfort zones long term. In addition, Erdogan et al. (2010) argue that the better use of ICT will, in turn, require better education. Lifelong education, as well as self-directed learning, will be key in this regard. Technology up-skilling is always a challenge, but this is an emerging need, particularly among sub-contractors. Up-skilling will, however, be of least concern for the technology-savvy workface in the coming digital era.

The role of government and industry in promoting wider adoption

Moore and Benbasat (1991) emphasise the significance of adoption being compulsory or voluntary and includes another dimension called “voluntariness of use” to the Rogers (1995) model. According to Moore and Benbasat (1991), the organisation’s freedom of choice to reject or adopt the technology depends heavily on the degree of compulsion involved. If there is a corporate policy regarding technology innovation either mandating or discouraging a particular innovation, such a policy takes the freedom of choice of rejection or adoption out of individuals’ hands. Supported by this theory, researchers have argued that innovation adoption in AEC industry (e.g. BIM) can be fostered by national and state-level policies (Manley and Mcfallan, 2006; Edirisinghe and London, 2015). In contrast, Loosemore’s (2015) research argues for industry-led rather than regulation-led innovation. Rewards (Gu and London, 2015) and subsidies (Singapore Government, 2015) are also considered as influential factors.

While the debate on the impact of policy on innovation is ongoing, there are examples of strong policies that mandated the use of technology in the construction industry (BIM policy in the UK and Singapore). A study on organisational motivation to adopt cloud computing by Marston et al. (2011) reported that compliance with regulations is an influential factor that can make firms reluctant to adopt technology. The lack of standards was reported to be a serious obstruction to decisions to adopt new technology for which the industry and academia both have a great deal of responsibility.