In the course of researching the article, “AI in AEC Updates, 2022,” published in February, I found that the construction stage of AEC had the largest number of applications using AI (Artificial Intelligence) — they were being used to handle a variety of tasks such as project planning, scheduling, progress monitoring, quality control, jobsite automation, and improving jobsite safety. Earlier this month, I was able to learn about the actual implementation of AI technology at a construction firm through an on-demand ENR webinar. The firm was Gilbane Building Company, a large integrated construction and facility management services firms with locations and projects worldwide, and the AI technology it is using comes from a London-based construction technology startup called Disperse.
In the webinar, the Gilbane team shared how this came about. In 2019, Gilbane was approached by one of its existing clients about to embark on a major new project (Figure 1), one for which they wanted to change the status quo. They wanted Gilbane to evaluate technology solutions that could improve not just construction efficiency and productivity but also create transparency for the many stakeholders that were involved in the project. A big ask was the ability to see potential problems as early as possible, so that they could be addressed to minimize delays and impact the schedule. Gilbane looked at several technologies — including some of those discussed in my earlier AI article — and finally decided on Disperse, after visiting projects in London where the solution was being piloted by large-scale contractors such as Mace and Canary Wharf Contractors (Figure 2).
At its core, Disperse is a building productivity solution that is designed to help construction companies track progress across their projects to identify inefficiencies and spot issues as soon as they occur, before they escalate into bigger, more expensive, problems. One of the fundamental tools it uses for this is capturing site photographs on a weekly basis, which are then processed with computer vision algorithms and integrated with the CAD drawings, BIM models, and construction schedules of the project to provide a visual record of the construction at any point of time. Also, because the visual data captured by the site photographs is mapped against the actual building element data coming from CAD and BIM, it is “building-aware” rather than being just a series of images — it “knows” about the elements that are being constructed, such as a beam, column, wall, duct, etc.
In addition, the integration of this building-aware construction-progress imagery with the building schedule allows the site progress at any time to be evaluated against what has been planned for that stage, allowing Disperse to generate progress reports on a weekly basis that can be shared with the construction team as well as the project stakeholders to keep them posted (Figure 4). The progress can also be consolidated into management dashboards at any time (Figure 5).
The integration with the project schedule allows Disperse to flag any problems or delays, so they can be dealt with by the project team sooner rather than later (Figure 6). Also, since Disperse can record construction progress throughout the life of the project, it provides a historical record of what was done when — an audit trail of sorts — which can be very helpful to resolve any disputes. You can literally peel back layers of time, so you can know, for example, what was behind a wall a few weeks ago.
In addition to the use of AI to stitch the individual site photographs into a digital 3D view of the construction site, AI is also being used to “recognize” the imagery that is being captured by processing it against the CAD drawings and BIM models and identifying the individual building elements that are being constructed. Disperse actually uses a mix of both AI and human expertise for this, as shown in Figure 7. This hybrid human-AI model is also being used to train the AI in construction so it can be continually improved. The human expertise comes from Disperse’s in-house team of architects and engineers,
In addition to the reporting, tracking, and productivity improvements that the Gilbane team found during their evaluation of Disperse, a major selling point also was that Disperse was committed to continually improving their product to make it work better for their customers. Thus, Gilbane would be in a position to help Disperse fine-tune its offering, making it, in turn, work better for Gilbane.
Gilbane’s initial deployment of Disperse that started in 2019 on one project (shown earlier in Figure 1) has now expanded to over 15 projects in New York, and it is continuing to grow. In addition to the transparency sought by the client that was the initial impetus for looking at Disperse, what the project team at Gilbane appreciates the most are the weekly progress reports and schedule updates that are delivered directly to their inboxes, allowing them to immediately spot issues and take action, stay on top of the schedules as well as adjust them if needed, and communicate objectively with their clients and subcontractors. By providing an exact record of what has been completed, it removes the subjectivity regarding what has been completed and when it was completed, which is often subjected to disputes and litigation in construction. A summary of the Disperse process at Gilbane is shown in Figure 8.
Some examples of the use of Disperse at Gilbane are shown below. Figure 9 shows the progress tracking report for the construction of a curtain wall. Similar reports can be generated for all critical construction tasks, and because they are automatically sent out to everyone on the project team, they are “in your face,” so to say, with nothing left to question. Since these reports are generated weekly, the progress from one week to another on a construction task can be easily compared.
The “building-aware” photo-documentation also allows actual construction errors to be detected, such as the example shown in Figure 10, based on the Disperse’s knowledge of what the construction is supposed to look like at that stage, of what is supposed to be where. This alert actually comes to the building team in the form of an early detection warning, allowing them to investigate and resolve the issue as soon as it received the notification.
And finally, an example of a management dashboard culled from Disperse’s reports is shown in Figure 11, allowing the progress and trends across multiple projects to be accurately captured and summarized for executive action.
In 2021, Gilbane won ENR’s award for “New York Contractor of the Year”, citing its partnership with Disperse as a contributing factor in achieving excellence for its clients. Its work on AI using Disperse also won an internal “Innovator of the Year Award” within the company. Disperse’s hybrid human-AI model is a great example of how AI can be used most effectively in construction to build “faster, cheaper, and better,” without sacrificing the deep human expertise that is so essential to any human endeavor.
Lachmi Khemlani is founder and editor of AECbytes. She has a Ph.D. in Architecture from UC Berkeley, specializing in intelligent building modeling, and consults and writes on AEC technology. She can be reached at lachmi@aecbytes.com.
Have comments or feedback on this article? Visit its AECbytes blog posting to share them with other readers or see what others have to say.
AECbytes content should not be reproduced on any other website, blog, print publication, or newsletter without permission.
This article explores the application of AI (Artificial Intelligence) in tools across each of the three main processes in AEC: Design, Construction, and Operations/FM. Some of the applications discussed are TestFit, Reconstruct, and ALICE Technologies, along with Nearmap and BricsCAD BIM.
The use of AI (artificial intelligence) in AEC applications is starting to see some traction. This article provides an overview of the technology underlying AI so we have a better understanding of it and then compiles what we have so far in terms of the use of AI in AEC.
