Technology

Machine Learning Is Transforming the Construction Industry

With machine learning, construction is becoming smarter as contractors measure progress, track materials delivery, as well as understand movement of labor and materials on-site.
By Karthik Venkatasubramanian
September 29, 2021
Topics
Technology

As the engineering and construction industry continues down its journey of digitization, one area that is becoming top-of-mind for organizations is machine learning. The engineering industry has already begun to realize the benefits from integrating data across systems, but the full potential that predictive analytics can bring to the industry has yet to be recognized.

As the adoption of machine learning continues to gain traction across the industry, it’s hard not to get excited about the opportunity that it presents in reimagining every phase of the plan, build and operate stages of the asset life cycle. Looking at advancements in deep learning, and the commercialized use of unstructured data like images, videos and text, as well as the ability to process big data through distributed cloud computing, these benefits will make themselves more apparent in existing work practices.

Unlocking machine learning With Data

Any good machine learning strategy needs data to work. Luckily, many construction businesses have troves of rich data available to them, even though they might not be using it today. By tapping into historical data from completed projects, construction businesses can use these existing repositories to create, train, compare and validate machine learning models. Machine learning presents a great opportunity to breathe new life into data archives by using them as training data for machine learning.

As additional aspects of the construction process become digitized, more data will be collected across different business functions and systems. This data can further improve the accuracy of machine learning models over time as they continue to learn from an organization’s projects and experiences.

Often termed feature selection, the wide array of data available from different systems can enable the identification of “markers,” which machine learning models can use to provide better insights to the successes or setbacks in a project. Developments in automated machine learning have greatly simplified and accelerated the process of identifying the right features, the correct model and its parameters for producing the most accurate insights.

machine learning Applications in Construction

Machine learning use cases are emerging daily, and what organizations care about most are the ones that create a positive impact on the bottom line of a project: on-time, on budget project delivery with zero safety incidents and no unexpected surprises that can cause delays and over-runs. Data-driven machine learning is making this goal easier to achieve by taking an analytical approach to project and risk management.

Project schedules can be complex and can comprise of hundreds or even thousands of activities. Using machine learning on completed projects can yield valuable insights into what worked and what did not. Machine learning is very efficient in recognizing patterns and can apply this recognition to current projects to make predictions through a process known as supervised learning.

Computer vision is another field of artificial intelligence being used to solve problems such as measuring progress on-site, tracking delivery of materials, ensuring compliance with social distancing rules and understanding movement of labor and materials on-site through the use of photos and videos. Before, it would have taken dozens of man hours to sort and analyze the unstructured data coming from videos and photos, but machine learning models using computer vision can do it in minutes, solving many problems that were difficult to solve previously. Machine learning can also be used to proactively manage budgets smartly by factoring change requests and variations that might be raised in the future.

Natural language processing, the ability for machine learning to read, understand and extract meaning from human language is another facet of technology being deployed in novel ways to reduce manual error, increase productivity and mitigate risks. NLP can spot areas of potential disputes and litigations through actively monitoring the data in project correspondence, review comments, field inspections and any other source of unstructured data. In addition to computer vision, NLP can also be used for early identification of potential health and safety risks. These systems often combine NLP with sentiment analysis to understand the subjective information embedded in the text and is often labelled as positive, negative or neutral.

Imagine an early warning system where concerns are identified as soon as they are mentioned in the systems tracking these items. Risk can immediately be surfaced around the quality of the material used, design changes not incorporated, electrical cords presenting a tripping hazard, improper use of fall arrest systems, use of portable ladders where not appropriate, exposed steel rebars, etc.

These machine learning driven techniques provide unparalleled opportunities to get ahead of the problem and take proactive measures to mitigate issues before they surface. These applications provide owners and contractors with the tools they need to better plan and respond to situations they might have otherwise not been prepared to handle.

The Takeaway

Whilst the use of machine learning is still in its infancy, the potential of machine learning using integrated data across systems is apparent. Predictions are only becoming more accurate as construction businesses connect data through a single platform and use this data across systems to provide more meaning and context to the insights presented.

Overall, machine learning is helping the construction sector become a lot smarter, plan better and deliver faster. Supply chain selection is also changing as subjective assessments are giving way to a more robust analytical way to select the right partners for the job. This is made possible by uniting data across disparate systems, allowing owners and contractors to see their subcontractor’s historical performance and using machine learning driven recommendations to select the best partner for the next job.

As far as the industry has come with the adoption of technology, it has only just scratched the surface of the transformative benefits of machine learning. As businesses become more fluent in utilizing their data, deploying machine learning based systems will become normalized. Machine learning based technologies are in use in so much of personal lives ranging from movie and shopping predictions to weather forecasts and voice-guided assistants—using these technologies in capital project delivery as they become more accessible will create efficiencies and drive businesses outcomes that will be transformative.

by Karthik Venkatasubramanian

Karthik Venkata Subramanian has global responsibility for defining and delivering the data strategy within Oracle Construction and Engineering. In a career spanning 20 years across four continents, he has been instrumental in building digital capabilities from the ground-up across many industries and organizations where he has successfully led several large-scale transformation programs that have been the featured in case studies, patent applications, industry publications and international conferences. He has an Engineering Degree in Electronics and Communication and a MBA from Melbourne Business School. Karthik is a passionate believer in creative innovation through the use of technology and infuses an element of “dataism” in almost everything he does. His current focus is on the application of data science techniques in solving real-world construction and engineering problems.

Related stories

Technology
Employing Supporting Roles for Your IT Team
By Christian Burger
For construction businesses to be effective in selecting, managing and deploying technology—especially when the influence, intelligence and complexity of that technology is growing—they need a new approach to IT.
Technology
Integrating Software and Hardware Technology in the Field
By Bryan Williams
Field technology has advanced increasingly in recent years. Combing the advancing software with hardware in the field can significantly improver performance on the jobsite.
Technology
Simplifying and Extending a Building's Lifecycle With Digital-Twin Technology
By César Flores Rodríguez
Digital-twin technology takes data beyond BIM, out of silos and into the interactive real world in real time.

Follow us




Subscribe to Our Newsletter

Stay in the know with the latest industry news, technology and our weekly features. Get early access to any CE events and webinars.