Can AI Improve Construction Efficiency? A Thorough Examination! Key Points and Cautions for On-site Implementation

Basic Knowledge of AI

AI (artificial intelligence), which is attracting attention in the construction industry as well, should first be understood at a basic level. AI is a general term for technologies that mimic and realize human intellectual work with computers. In recent years, advances in machine learning and deep learning have enabled the analysis of vast amounts of data from images and sensors, allowing systems to learn patterns and rules and make sophisticated decisions. For example, recognizing objects in photos or predicting the future from past data can now be done quickly and with high accuracy. In the construction field, initiatives to use these AI technologies to automate and improve the efficiency of various on-site tasks are progressing. Inspections and planning that used to rely on human experience and intuition are increasingly being supported objectively by AI based on data. Such AI utilization is expected to be a pillar supporting the construction industry’s DX (digital transformation).

Benefits and Effects of On-site Implementation

For construction sites facing numerous issues such as severe labor shortages, long working hours, and low productivity, introducing AI technology can be a trump card for solving problems. By applying AI to on-site operations, the following main benefits and effects can be expected.

• Improved operational efficiency and time reduction: Automating tasks such as surveying and inspections allows operations to be completed in less time than before. By reducing manual labor and compressing waiting times and redundant tasks, the overall schedule is sped up.

• Addressing labor shortages through manpower reduction: Against the backdrop of chronic shortages of skilled workers, AI can act as a “smart worker.” By having AI take over tasks that could not be covered by a small workforce, operations can be carried out with fewer personnel. Also, by learning and sharing the knowledge of seasoned workers, AI helps with skill succession.

• Improved safety: Assigning dangerous tasks to AI and machines can reduce the risk of occupational accidents. For example, having drones and AI perform high-altitude inspections eliminates the need for people to enter hazardous areas. Real-time video AI monitoring can detect signs of accidents early and issue warnings.

• Standardization and improvement of quality: Using AI in inspection processes eliminates human variability and enables checks based on uniform quality standards. AI can detect even fine defects, reducing rework and raising construction quality, which in turn contributes to increased customer satisfaction.

• Cost reduction: The above effects—efficiency improvements, manpower reduction, and enhanced safety and quality—can lead to overall cost savings. Reducing labor costs through shorter work times, avoiding wasteful expenses from fewer mistakes and accidents, and minimizing material losses through proper management all mean that AI implementation can generate high long-term return on investment.

Main On-site Work Areas Where AI Can Improve Efficiency

Now let’s look specifically at which on-site tasks AI can help streamline in construction. From surveying to safety management, AI utilization is advancing across a wide range of fields. We will examine how AI is useful in each area and the current effects.

Surveying (Grasping Current Conditions)

Surveying is one of the areas where AI utilization has particularly advanced. Topographic surveying, which traditionally involved manually setting up a total station and measuring point by point, can be greatly streamlined by combining drones and image analysis AI. Drones photograph the site from above, and AI analyzes those images to automatically generate 3D point cloud data. This allows wide areas of land elevations and shapes to be grasped in a short time; for example, in a forestry development site, point cloud data for the entire terrain can be obtained from the air in a matter of tens of minutes. There is no longer a need for people to climb dangerous slopes or spend days surveying, resulting in a dramatic improvement in safety and work efficiency. From the acquired 3D survey data, heights and distances of arbitrary cross-sections can be measured later as needed, so AI can automatically perform as-built checks and earthwork volume calculations (calculating cut-and-fill volumes). The Ministry of Land, Infrastructure, Transport and Tourism’s promoted i-Construction also places 3D surveying as a core element, and surveying technologies using AI have already been put into practical use at many sites.

As-built Management

As-built management is the process of confirming whether completed structures and terrain match the shapes and dimensions in the design drawings. AI also demonstrates power in this area. For example, after embankment or paving in road construction, survey staff traditionally measured heights at many locations to check the as-built condition, but now 3D scanners or drone photography combined with AI analysis can digitally capture the entire surface shape. AI compares the design data with the as-built point cloud to automatically perform as-built inspection, visualizing areas of excess or deficiency with color coding. This instantly reveals locations that require rework, leading to quality assurance and reduced rework. Since as-built records can be stored as photos or point cloud data, paperwork is reduced and reporting to clients is smoother. AI-driven as-built management has progressed particularly in civil engineering and achieves significant efficiency gains on sites handling large volumes of earthwork.

Safety Management

AI utilization is also attracting attention in construction site safety management. With AI-equipped surveillance cameras, sites can be monitored 24/7, detecting dangerous situations and issuing alarms. For example, AI can check from camera footage whether workers are correctly wearing helmets and safety harnesses, and immediately notify the site supervisor if someone is not wearing them. Systems that sound alarms when people or heavy machinery threaten to enter restricted areas have also been commercialized. Research is progressing on monitoring workers’ movements and behavior to detect signs of accidents such as slips or falls. These AI-based real-time monitoring systems can cover moments that human supervisors might miss and are expected to reduce oversight of near-miss incidents (close calls).

Meanwhile, AI is beginning to be installed on heavy machinery as well, and safety features that automatically detect and stop excavators (backhoes) and cranes when people or obstacles are nearby have appeared. AI-driven safety management can greatly reduce “I didn’t notice” and “careless mistakes,” bringing sites closer to zero-accident environments.

Schedule and Progress Management

Schedule (timeline) management is extremely important in large-scale construction projects, and AI shows strength in this area as well. First, there are technologies that automatically grasp on-site progress. For example, attempts are underway to have AI analyze daily images taken by fixed cameras or drones to determine the progress (how much has been completed) of buildings and infrastructure. This allows AI to objectively calculate “X% of the plan completed” without people visually estimating the progress rate. Moreover, AI trained on numerous past project datasets can predict the risk of schedule delays early based on the current pace and resources deployed. If AI indicates “At this rate, the final phase may be delayed by Y days,” countermeasures can be taken in advance. Going a step further, AI is beginning to be used to optimize schedule planning. Systems in which AI simulates construction procedures and equipment placement to present the fastest-completion plan are emerging. Although still implemented only in some cases, in the future it may become common to base construction plans on AI-generated schedules. In any case, AI utilization in progress and schedule management supports waste-free planning and rapid decision-making, ultimately leading to shorter schedules and cost reductions.

Material Management

AI is also useful for ordering and inventory management of materials used on site. Construction work requires properly arranging and managing various materials—from concrete and rebar to bolts and fuel for heavy machinery. Introducing AI can predict the required amount of materials based on past construction data and current progress, preventing stockouts and overstocking. For example, one AI system analyzes data from site sensors and daily reports and predicts “rebar will be insufficient in 3 days,” prompting advance reorder. This reduces the risk of work stoppages due to material shortages. There are also cases where image-recognition AI automatically counts components in material yards or streamlines incoming inspection. Further, AI that learns price fluctuation data is being researched to suggest the optimal purchase timing. AI in material management not only supports smooth on-site progress but also contributes to cost compression by avoiding excess inventory.

Quality Inspection

AI is active in post-construction quality inspections and finish checks as well. Traditionally, concrete cracking and surface finish irregularities relied on the eyes of veteran inspectors, but image-recognition AI now enables automated inspections. For example, systems exist that photograph wall tiles or painted surfaces and have AI detect defective areas. AI can identify minute defects that human eyes might overlook with high precision, enabling early correction of quality issues.

Major general contractors have also developed technologies that automate rebar inspection by detecting the number and spacing of rebars with AI. This has dramatically shortened inspection times and improved inspection accuracy.

Furthermore, advanced initiatives combining AI, 3D scanning, and AR (augmented reality) have begun. When a structure is scanned with a smartphone or tablet, AI judges defect locations and displays markings in the real space via AR. Inspectors can directly correct the indicated areas, reducing missed rework. Such advances in AI-driven quality inspection are expected to standardize and speed up inspection processes. In the future, AI will likely become the “inspector’s partner,” constantly checking sites and serving as the last line of defense for quality assurance.

Examples of AI Utilization

Here are several examples of actual cases where AI has produced results.

• Survey efficiency improvement with drone × AI (Obayashi Corporation): Obayashi Corporation introduced drone and AI analysis for surveying in civil engineering works. They built a system that automatically generates orthoimages and point cloud models in the cloud from drone photography, greatly simplifying data processing that previously required dedicated technicians. As a result, surveying work time was reduced to a fraction of what it used to be, enabling accurate current-condition data to be obtained on the same day. Data sharing is also smooth on the cloud, so all stakeholders can grasp the latest terrain information in real time.

• Automation of rebar inspection using AI (Obayashi Corporation): Obayashi Corporation developed a system that automates rebar inspection for reinforced concrete structures using AI. AI analyzes photos of rebars taken on site and automatically measures and judges the number, diameter, and spacing. The accuracy has reached a level comparable to experienced inspectors, successfully reducing inspection time significantly. Because AI judgment results are visualized, site supervisors can efficiently confirm corrections. Rebar inspections that used to take half a day can now be completed in a short time, reducing variability in quality.

• Safety monitoring with AI cameras (Taisei Corporation): Taisei Corporation prototyped a “smart helmet” that combines a 360-degree camera mounted on workers’ helmets with AI image analysis to manage site safety and grasp progress. The AI analyzes footage in real time to detect dangerous behavior and automatically check work-area progress. If abnormalities are detected, notifications are immediately sent to managers, allowing remote site monitoring. This is expected to reduce the labor required for safety patrols and reduce human error. By accumulating and analyzing collected site footage data, the system also aims to contribute to future prediction of signs of occupational accidents.

Although the above are examples from major companies, recently small and medium construction firms have also started using cloud services and smart construction systems with built-in AI. With support from the Ministry of Land, Infrastructure, Transport and Tourism, local construction sites are adopting drone surveying and automatic control of heavy machinery, producing cases that achieve productivity improvements of more than 20%. Across the construction industry, a wave of DX centered on AI is steadily spreading.

Points to Note When Introducing AI

When introducing AI on-site, it is not enough to buy tools indiscriminately. To maximize effectiveness, pay attention to the following points.

• Clarify objectives and issues: First, clearly define which problems at your site you want to solve with AI. If you introduce technology just because it is “the latest,” it will not take root on site. Identify specific issues such as “surveying takes too long” or “many mistakes occur due to human error,” and select AI solutions suited to those issues.

• Build consensus on-site: It is also important to gain the understanding and cooperation of the people who will actually use the system. To reduce resistance to new systems, hold briefings and demos for site staff before implementation to share benefits and usage. If you can customize the system while incorporating on-site feedback, the site will proactively engage with it.

• Start small with tests: Rather than full-scale implementation at all sites immediately, it is recommended to start with pilot sites or specific processes. By validating effects and issues in a pilot project and addressing problems before full deployment, you can reduce the risk of failure.

• Prepare data and environment: Data is the lifeblood of AI. If you introduce photo AI, set shooting rules to ensure image quality; if you use sensors, improve communication environments—such preparation is key to success. If learning data such as past drawings and construction records are required, organize and digitize them in advance.

• Define operational rules and follow-up: You must also decide the operational flow after the system is installed. For example, determine in advance who will handle AI warnings and how, who will check the data, and so on. In the initial phase after introduction, assign personnel to answer on-site questions and promptly handle defects; such a follow-up system is essential. Actively utilize vendor support as well.

• Consider legal regulations and privacy: As drone flights may require permission, the use of new technologies must comply with relevant laws and regulations. In systems that constantly monitor with cameras, privacy considerations are also important. When introducing systems, thoroughly consider legal compliance and ethical aspects.

Technical and Human Hurdles

Be aware of the technical and human challenges commonly faced when promoting AI adoption.

• Cost barriers: AI implementation involves initial and running costs. Equipment purchases, software subscription fees, and employee training can be burdensome for small and medium enterprises. However, inexpensive trial solutions have become more available due to cloud service proliferation, and national subsidy programs can be leveraged. It is necessary to estimate cost-effectiveness and clarify long-term benefits and payback prospects.

• Lack of personnel and skills: Many on-site personnel are not familiar with digital technologies, and a shortage of DX talent capable of using AI is an issue. There may be little time available for training. Consider initiatives to raise IT literacy within the company or bringing in external specialists. Also, to have AI learn the expertise of veteran workers, efforts to formalize tacit knowledge are required.

• On-site environment and infrastructure: Construction sites may have unstable communications or difficulty securing power. If using cloud AI, you need to prepare the site’s communication environment, and handling large volumes of data requires high-performance PCs or tablets. These infrastructure improvements can take time and money.

• Integration with existing systems: AI tools alone may not maximize their benefits. Many cases require linkage with existing construction management systems, accounting systems, and BIM/CIM data to be truly useful. System integration requires technical adjustments and customization, which can be a hurdle for companies with limited IT expertise. Consult vendors and proceed with integration step by step.

• Organizational culture change: Finally, there is the hurdle of changing human attitudes. The construction industry has long-established customs and craftsman culture, and there is psychological resistance to new technologies. It takes time to dispel pride in “we’ve always done it this way” and vague anxieties about AI. Management must play the role of championing DX and cultivate a new culture together with on-site personnel. Foster an environment that tolerates failure while celebrating challenges, and promote DX from both top-down and bottom-up directions.

Using Smartphone + GNSS for Simple Surveying “LRTK” as the First Step Toward On-site DX

Finally, we introduce LRTK, a solution highly compatible with AI and ideal as an entry point for on-site DX. LRTK (LRTK) is a system that combines a smartphone with high-precision GNSS and is a simple surveying tool that allows anyone to easily achieve centimeter-class positioning (half-inch accuracy). Surveying work that traditionally required specialized equipment and skilled technicians can be performed with high precision through intuitive smartphone operation using LRTK. For example, site surveys that used to take several people half a day can be completed quickly by one person with LRTK, contributing significantly to time savings and personnel reductions.

Position and photo data acquired with LRTK can be directly used for AI-based point cloud analysis and AR display. If AI analyzes high-precision 3D survey data obtained by smartphone, automatic as-built judgment and quantity calculation can be performed instantly. Also, because LRTK provides accurately linked position information, AR displays that overlay design models onto the real world on smartphones and tablets can be performed with high precision. For example, at a construction site, you can view a projected completion image or inspection points overlaid on the real scene through a smartphone screen, enabling intuitive and easy-to-understand site management.

While LRTK itself is an excellent survey DX tool, it yields even greater effects when combined with AI technology, point cloud processing, and AR visualization. Initial costs are lower than traditional surveying equipment, and because LRTK is an i-Construction–compliant product supported by the Ministry of Land, Infrastructure, Transport and Tourism, it is reliable. As a first step in on-site DX, starting with digital surveying using LRTK allows you to smoothly connect the obtained data to AI analysis and other DX measures. LRTK can be a strong partner in realizing smart construction management suited to the AI era.