BIM & Artificial Intelligence
Updated: May 16, 2019
We all know pretty much about BIM by now. BIM over the last decade and a half has totally transformed the construction industry. The way construction industry worked during 1990’s and the way it works now is totally a contrast when one compares some aspects of designing, drawings generation, communication, collaboration etc. So where is the BIM heading to when the world of IT is heading towards AI. Is BIM AI or can it be AI?
So what is AI?
As per the definition on Wikipedia
“Artificial intelligence (AI, also machine intelligence, MI) is intelligence exhibited by machines, rather than humans or other animals (natural intelligence, NI). In computer science, the field of AI research defines itself as the study of "intelligent agents": any device that perceives its environment and takes actions that maximize its chance of success at some goal. Colloquially, the term "artificial intelligence" is applied when a machine mimics "cognitive" functions that humans associate with other human minds, such as "learning" and "problem solving"
From this statement, AI involves cognitive functions similar to that of human minds. Cognition means – “the mental action or process of acquiring knowledge and understanding through thought, experience, and the senses.”
So, does BIM have any of these qualities to satisfy the claim an AI?
BIM being a process by itself cannot qualify to be an AI process as a whole. However, BIM within itself has many components that are AI, maturing to be AI, contributing to AI and can become a total AI.
A few observations on the existing AI in the construction industry through BIM.
1. Probably most of us know of the user gravity controls, of a very useful aspect of popular BIM Walkthrough/Visualizations Tools like Navisworks, BIMx etc.,. This tool can detect the building horizontal components and give the user the feel of standing on ground and realistic eye-view sight. There have been many advancements in this recently, wherein the tools can simulate the movement on ramps and stairs. Now that we have been seeing this over a monitor for a while now, we might not have found it great. The same, when seen through AR & VR (Augmented & Virtual Reality) applications, can simulate the experience of being on AI. Imagine sliding on a huge ramp under construction wearing AR google and watch the yet to be built Virtual Building. It sounds awesome. Yes probably just a few years away when this might be a reality.
2. Few years before, most of the BIM tools were redundant/unusable without some simple plugins. Some notable plugins – Element Auto-Numbering Plugins, Hanger Plugins, Duct/Pipe Slicing Plugins, Sprinkler Placement etc. All these plugins, on the backend, knew the logic of the tasks they perform. They are in fact similar to a human in problem-solving related to placement of objects; had they any obstacles on their way in form of clashes, overlaps, design rules, constructability issues etc., using a predefined written code of solutions they can resolve the issues. The major aspect that’s missing is that of machine learning where the machine learns from its tasks.
In some of the recent improvements on the BIM Softwares especially those are on the cloud, we can see that software are capable of providing multiple solutions, followed by comparison and impact matrices. Developers of these are monitoring the usage very frequently and are trying to make them more intelligent every day. Probably a right move towards automating the machine learning to complete the AI. Further one needs to note that a number of Foundation Rules govern the BIM process in the form of codes, statutory requirements etc. forming a standard rule that has very limited scope for deviating. Considering this limited scope and the flexibility of tools available, these plugins are surely bringing in a solution similar to AI.
3. Let’s compare quantification between CAD and BIM. Surely every CAD software can quantify everything they are equipped with to draw. While CAD provides dumb length, area, and volumes of everything, BIM can provide a selective output depending on the element/material selected – volumes for concrete, lengths & weights for steel, areas for masonry, lengths & dis for pipe, duct, conduit etc.
Surely indeed, they are AI, but an informed and preprogrammed AI that has scope to be detailed out to be more flexible, resourceful and productive.
For instance, a software named Wood Workshop knows all the science behind wood – color, shades, grains, look, feel, layout etc. A user using this can get a very realistic output of any kind of wood they are looking for, just by knowing the color shades and grains pattern. Imagine your BIM software doing the same sometime in future where the user is not let to crib “this software should have had this tool; how could they not provide this basic tool?”
4. Now a look at constructability check tools like Solibri or Revit Model Review. These tools are capable of checking any given element against set rules around other variety of elements. One needs to understand is that the user keys in rules and not layouts. So the software somewhere is able to make its own choices to arrive at a number of checks, elements to be checked against, rules to be checked for and so on. Software code might be in-built for it. However, we need to acknowledge that the building designs are being rarely similar and not in many instances, these tools have completely failed in achieving their basic goals. Hence, probably the code of these tools are already a miniature AI.
5. The emerging world of photogrammetry, wherein the scanners are capable of detecting hard surfaces and the length based on signal beaming etc. is been used very extensively in the construction industry. In addition, there are attempts by various companies to identify differences between scan data and the virtual model to automate the process. Tools like Autodesk Remake, Bentley Context Capture etc. are intelligent enough to create a model from photographs. There is a further development here to detect the materials and accordingly model them as solids, mesh or surfaces. Tools like Scan to BIM from imaginit can differentiate, detect and draw on own from point cloud the walls, pipes & ducts. Surely a guided move towards AI.
Check out a list of photogrammetry tools available at this Wikipedia link - https://en.wikipedia.org/wiki/Comparison_of_photogrammetry_software
6. A lot of AI is happening behind the curtains on BIM Documentation. Tools like BIM 360 Docs, Project Wise and Bluebeam can read through the pdf's and editable drawings, search, relate and associate with corresponding views. Like the section & elevation markers on plans are capable of relating themselves to section or elevation views. The ongoing development here includes adding abilities to detect corresponding views across the construction disciplines. Like architectural section detecting and relating itself to the piping section at the same location or orientation on a completely different drawing set with probably a different title block as well. A true AI being coded into software that has on backend vast learning relating to graphics, drawings, systems, processes and surely the human thinking.
7. Now the entire idea of BIM Building Performance Studies etc. is an AI. Building Performance Studies involves an understanding of a lot of user & building operation profiles from the known world, followed by interpolating of these values against future predictions, cross checks against standard operation manuals and so on. This software is programmed to analyze, organize, detect, predict and solve some of the unknown problems (to be seen in the future). Guess how can one contradict that Building Performance is not AI.
8. Some of the building safety analysis related to Seismic and Fire are superb renderings of real-life scenarios that might or might not happen. Probably an unnoticed AI that has been in use for decades.
9. The world of infrastructure has many more examples of AI in the form of hydrology studies of reservoir backwaters, flood simulations, traffic analysis, high altitude wind analysis, physical deteriorations due to environmental factors etc. Every process/activity/study is a very unique AI each of them.
10. Last but not least, a new industry of demolition engineering is very active and proceeding towards maturity. We need to note that the software can predict and demonstrate the building collapse depending on the location of explosive. To be able to predict this, the software should surely know and predict accurately the structure of the building from the known information or point cloud; conduct a lot of iterations & predictions and suggest the best solution to demolition a building.
To conclude, AI is already there in BIM Industry in a very different form. Unlike AI in other industries, in BIM, it is more informed, predefined and pre-coded because of the vast construction knowledge database of diversity in designing, diversity in practice, building physics, building mechanics, building sciences and their systematic comparisons available to the software developers. Construction Industry is truly one of its kind of industries, that adopts silently to any of the emerging technologies be it physical mechanics or IT innovations.
So BIM indeed has started adopting AI in its own way. Might be a long way and hundreds of debates before concluding. No wonder if BIM ends up giving rise a very new definition or a new technology within AI to suit its requirements. Construction Industry surely needs AI in its own form.
Mr. Arunkumar is a BIM Evangelist from Bangalore, India, having few years of BIM Experience on projects, standardization, and implementation. His common interests include current affairs, technology exploration, market study, lateral thinking and discussion within his close circles on a variety of topics ranging from professional to personal subjects.