This is a follow-up to my first look at the application of AI (Artificial Intelligence) technology to the AEC industry that was published in February 2019 — the article, “AI in AEC: An Introduction.” While most of that article was devoted to discussing the technology underlying AI and its broader implementation, it did highlight the three AEC-specific commercially available applications I had found up until then that were using AI: BASIS, a project planning tool that had just been acquired by InEight; BricsCAD BIM, a BIM application that was using AI for automatic element classification; and OpenSpace, a construction progress tracking tool that was using AI to automatically stitch together photos and videos of the construction site (Figure 1). Towards the end of that article, I anticipated that we would see a lot more of AI in AEC going forward, both in the form of new applications as well as in the form of enhanced features in existing applications.
Three years down the road, how far have we come? That is why I set out to investigate, and my findings are captured in this article. It looks at the application of AI in tools across each of the three main processes in AEC: Design, Construction, and Operations/FM.
In the “AI in AEC: An Introduction” article, one of the three applications using AI that was highlighted was BricsCAD BIM. A subsequent article on the BricsCAD 2021 Summit that was recently published described this in more detail, showing an AI-based tool called BIMify in BricsCAD BIM. This tool uses machine learning algorithms to analyze the model, automatically organize and classify its elements, and assign attributes to them, without needing them to be laboriously defined by the user.
Another example of the use of AI for design — of buildings as well as infrastructure — that was described in a recent AECbytes article was in my review of Nearmap, which provides aerial imagery for AEC. For the past couple of years, Nearmap has been working on using AI technology to make sense of the captured data by automatically identifying specific features in the imagery and making them available in the form of “AI Packs” for different features such as roads, houses, roofs, vegetation, construction sites, and so on. This AI-extracted information allows customers in different industries to easily access the data from the imagery that is most relevant to them (Figure 2).
While the attempt to develop automation in design tools is not new — see, for example, the articles, “Bluethink House Designer: Automating the Re-use of Design Knowledge” and “SITEOPS: Applying Optimization Technology to Site Design,” published in 2009 and 2008 respectively — the trend has accelerated with advancements in computing technology, and many firms are now developing such tools in-house. Recent examples that were highlighted in AECbytes include Bryden Wood and Thornton Tomasseti. In particular, the article on Thornton Tomasseti described an automated design tool called Asterisk, which can generate multiple conceptual design options for a structure based on specific parameters, allowing the designer to explore multiple options. Asterisk has an integration with a separate emerging product called TestFit, which uses AI technology to streamline design tasks in the form of a “building configurator.” TestFit reacts dynamically to the user’s inputs, either through input parameters (such as residential, unit mix, etc.) or spatial constraints (such as zoning, setback, easement, etc.), and produces a building configuration that meets those input requirements in real time (Figure 3). While TestFit does not use machine learning trained by large datasets to power its AI, it does provide the “intelligence” that can dramatically reduce design time from weeks to minutes.
The construction stage of AEC currently has the largest number of applications using AI. Several of these are in the reality capture space, similar to the OpenSpace application I had highlighted in my 2019 Intro article. There is Reconstruct, a remote quality control and progress monitoring solution for construction projects that can combine reality capture data, drawings, models, and the project schedule to show an overlay of what is built versus what is planned, identifying potential conflicts and areas that are at risk of delay (Figure 4). It uses AI technology in three ways: to stitch together the thousands of captured images and videos into measurable 3D reality models, floor plans and site maps; align these as-built models, floor plans, and maps with the 2D and 3D design models of the project, providing a visual overlay of design versus reality; and analyze the relationship between the reality model, BIM, and schedule to produce actionable insight about progress deviations and identify risk for schedule delays.
Two other applications that also use AI for stitching together reality capture data into 3D models that can be compared with the actual BIM models of the project to track construction progress and monitor the site for errors are Buildots and Avvir. In the case of Buildots, the images/videos are captured with hardhat-mounted 360-degree cameras (similar to OpenSpace), while for Avvir, the reality data can also include point clouds from laser scans. All of these companies have attracted substantial venture funding, showing that investors view this area of “AI in AEC” as having tremendous potential to help construction firms build projects faster, cheaper, and better. (See the news about the investments into Buildots, Avvir, Reconstruct, and OpenSpace.)
One of the earliest examples of the use of AI for any aspect of AEC that I was made aware of comes from a company called ALICE Technologies, which I had actually visited several years ago when it was still in development at a startup tech incubator called Y Combinator. Based on research conducted at Stanford University, ALICE applies AI to automate construction scheduling. It allows construction companies to create “recipes” for different construction tasks – for example, the recipe for pouring a concrete slab would include labor, materials, duration, equipment, rates, what needs to be completed prior to installation, etc. These recipes are then juxtaposed with the resources that are available for the project, yielding different construction scenarios that can be explored, each with its own schedule and estimate. (Figure 5). Once a particular plan has been selected, it can be further fine-tuned and managed for the duration of the project.
Some other construction applications using AI that I came across while researching this article include Kwant, which uses AI to process jobsite data collected in real time by sensors and provide actionable analytics like schedule and cost risk, optimize workforce, and predict and prevent safety incidents; INTSITE, which uses AI to optimize and automate the operation of construction cranes and other heavy machinery at a jobsite; and Newmetrix, which captures jobsite data with mobile devices and processes them to measure, predict, and prevent safety incidents.
The AI-powered project planning application, BASIS, that I had written about in my 2019 article has been rebranded as InEight Schedule. I also came across this news of a German startup, Conxai, which is working on a “no-code AI platform” for the construction industry, but no additional details were provided about what exactly it will do.
And finally, I was heartened to read this article about AI waste-sorting robots being deployed at a newly-opened recycling plant in Finland to capture valuable materials from construction waste streams. It is good to use AI smarts being deployed not just to build, but also to re-use and recycle.
While I did not find any substantial developments yet in the application of AI for the operation and maintenance of buildings and infrastructure, it is definitely being envisioned, as in this article which discusses how machine learning can change city design and enable the creation of more sustainable, liveable cities. Given the vast amount of data that has to be analyzed for any aspect of city design, it seems ripe for the application of machine learning. And given the global aspiration for “smart cities,” it seems inevitable that AI technology will be deployed in this space sooner rather than later.
On the level of individual buildings, I did come across a platform called Hank that works with building controls and uses AI to manage HVAC energy, comfort and air quality. Hank has just been acquired by a US-based real estate and investment management firm, JLL, which will hopefully spur the development of more such tools that use AI for improved operation and maintenance.
Towards the end of 2021, just a couple of months ago, I came across a fascinating article in the New York Times Book Review called, “A Robot Wrote This Review,” and I actually tried out the A.I. writing program, Sudowrite, that it referenced. While I did not use that program to write any portion of this article (!), I was extremely impressed at how good it was, and it motivated me to see how far AI has progressed in AEC since I first looked at it in 2019.
While there hasn’t quite been the tsunami of applications that I had expected, the ones that have been developed and are commercially available are very promising. And perhaps, not having a deluge of AI applications is a sign that we are not falling for the hype surrounding AI. All the applications I saw are meaningful and help to tackle genuine pain points in the AEC industry. Hopefully, the use of AI will continue to expand to solve the thornier problems we face in the design, construction, and operation of buildings and infrastructure.
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 firstname.lastname@example.org.
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.
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.
This article explores the cutting-edge AEC technology applications developed by CORE Studio, a dedicated software development group at the leading structural engineering firm, Thornton Tomasetti.
This article explores the rule-based design technology of Bluethink House Designer, developed for one of the largest home building companies in Norway. It looks at how this application automatically applies embedded knowledge to support dramatic improvements in design and construction.
This article looks at the application of evolutionary computing technology -- which uses optimization to find the most effective solutions to a problem given a set of constraints -- to the field of site design and land development in SITEOPS.