Artificial Intelligence and Machine Learning in the AEC Industry

Although it may still seem like a distant concept to some, Artificial Intelligence (AI) is much more present in our daily lives than we imagine. As a branch of computer science dedicated to the simulation of human brain processes and behaviors in machines, this technological revolution is also present in the construction sector, from a project’s early stages to the verification of errors and inconsistencies at the construction site. Artificial intelligence can be especially useful for performing repetitive and analytical tasks, proving to be incredibly beneficial to the sector. In line with these trends, RIMOND has been following these advances and has actively sought to apply artificial intelligence and machine learning to maximize our excellence and service efficiency to our recent projects on various programs and scales.

Bringing countless possibilities to architectural projects

In architecture, the design process requires multiple decision making, which in turn has implications on a variety of other decisions that can extend infinitely, therefore becoming an exercise of choices and concessions. For example, in order to use structural modulation more efficiently, utilize the maximum amount of an area in accordance with local urban legislation, or improve natural lighting within a building, it is necessary to go through numerous iterations before reaching the most appropriate result. In these examples, artificial intelligence allows us to take advantage of what computers are best at: accessing an unlimited amount of data and bringing an array of possible solutions to the most diverse problems. One example is the concept of generative design, which unites parametric design and artificial intelligence to process project data added by the architect. It is based on the exhaustive exploration of formal alternatives, derived from certain assumptions defined by the designer for a certain purpose.

But AI can be integrated into architecture even before the first idea or drawing is scribbled on a piece of paper. Dall-e, for example, is a tool that promises to transform any textual information into an image. The user types what they imagine and the computer creates several possible images and even edits existing images. Dall-e seeks standards by analyzing millions of digital images, as well as text subtitles that describe what each image represents. This way, it learns to recognize the links between images and words. In other words, this software can revolutionize the processes of project representation.

Making the construction site more efficient and intelligent

Artificial intelligence can also play an important role in the construction of buildings and infrastructure, with the potential to reduce waste, predict possible issues, coordinate off-site manufacturing and, most importantly, make the construction site safer. When combined with machine learning, the possibilities and advantages become even greater. Through sensors and cameras, machines are able to analyze themselves, evaluate the efficiency of the processes and refine each one of them. Data collected by all robots involved in production can be gathered and examined by a central artificial intelligence system, which learns which problems are most likely to happen. With machine learning, this central artificial intelligence is also able to formulate solutions to problems, anticipating possible failures.

Another concept that is on the rise is that of gamification, which is about adding game mechanics, social engagement and contribution statistics to encourage an individual in environments such as websites, social networks and even a construction site. In the latter, it can be valuable for team training, security management or for the monitoring of construction and processes in order to avoid errors. The so-called digital twins have been developed to carry this out – a model that integrates real-time data of a built asset with its digital representation to create information during the project life cycle. 

For the construction of Al Wasl Plaza Trellis – one of the largest domes of its type in the world–designed by Adrian Smith + Gordon Gill Architects, solutions were developed to not only make the process possible, but also more efficient and intelligent. RIMOND’s Data-Driven Design Technology team, used an AI-based advanced form finding tool: a tailor-made algorithm that optimized the geometry of the dome to obtain the best alignment between structural modules of folded steel tubes through thousands of formal iterations. Alongside this, a neural network was trained to create ideal geometric configurations in compliance with the intention of intent and project manufacturing, in relation to structural flexion, cutting and welding. Furthermore, we have worked on a generative design solution which has automated the complex network of MEP cable routing inside the steel pipes forming the dome, that offers the world’s largest immersive experience. Thanks to these winning solutions, we were able to increase the pace of design, accelerating both the fabrication and construction processes.

Lorenzo Pirone, Design Technology Solution Group Lead at RIMOND, regarding future applications of AI in the AEC industry, says that “We have been recently working on how we can improve the design process of modularity in building projects using generative design applications based on AI that can learn from the constraints of materials, fabrication, logistics, environment, and urban legislations. In construction, there are a lot of industry-standards workflows and tasks that can be streamlined using AI. The data which could be collected using IoT devices on site could be used to instruct a large neural network to perform a variety of reporting tasks that would improve safety, quality and reduce times in projects.”

Even though all of these possibilities are starting to appear in different areas, they are all very new concepts — especially in the construction industry — that still need to mature to reach a wider market. Other concepts, such as “big data” and “internet of things”, have already begun to become part of our vocabulary, demonstrating how quickly these changes occur. Despite having almost unimaginable capabilities, it is essential that machines and algorithms continue to receive appropriate input from humans, in order to be well directed and bring positive results. Computers can help organize and prioritize decisions, but only humans — at least for now — can decide what is important and what is not.