Building services engineer here. I think AR could be useful for "x-ray" views of services above ceilings and in risers where you're trying to locate concealed equipment. When it's sitting in front of you in a plantroom you really just need to look at it or have a regular screen to tell you what's going on in terms of operations.
The big assumption is that the 3D model of equipment used by an AR app matches what's actually installed on site. This is rare in my experience - particularly over time as maintenance is performed in bits and pieces without documentation being updated. It's exceedingly rare to find an old building with accurate documentation of where everything is. I'd love an app that could compare a laser scan of installed plant with the 3D design model and automatically align them into an "as-built".
It’s also impossible in the vast, vast majority of real-life situations.
As DeNiro said in the terrible
Movie “Ronin” — “But the map…the map is not the territory.”
The 3D model will differ from the actual construction. From the very beginning and more and more over time. Documentation is hard enough for software projects. Why would we expect documentation quality of construction to be better?
I can see this being used for retrofit designing. Use a LiDAR point cloud to create a model of an existing open facility, all the exposed machinery and piping. Then suggest changes for future installation and have current staff walk through the plant checking the virtual future piping/etc.
We already do that without the 3D glasses though. Just 3D rendered to computer screen.
The 3D model will differ from the actual construction. From the very beginning and more and more over time.
Not really. The point of construction is to turn the "3D model" into reality.
Some problems in the original planning can and will be spotted during construction but that just means plans are updated once conflicts are resolved.
There's even a occupation specialized in ensuring that what's being constructed complies with the "3D model": construction and building inspectors.
The mental models that are formed by software developers based on their personal experience within software development activities is not compatible with any day-to-day practices of any engineering field.
Documentation is hard enough for software projects. Why would we expect documentation quality of construction to be better?
Because it is, by it's very nature and purpose. Practically all of the engineering deliverables are written documents. These documents held engineers liable to any defect. In engineering, documents matter the most.
Engineering is not software development, and software development is not engineering. Engineers need to put in writing stuff that works for decades, at their financial and even civil and criminal risk. You don't just hit the "build" button and retry if something failed. You put into writing exactly how something needs to be done in order to meet all requirements, and make sure that everything is done according to what was put in writing. If anything goes wrong, what's written down is the engineer's only defense.
There's even a occupation specialized in ensuring that what's being constructed complies with the "3D model": construction and building inspectors.
Respectfully, I’ve done this specific job in the past for industrial/commercial building construction. It’s very, very difficult to get builders to build the thing to perfectly match the drawings. They often build what is convenient, or make absent-minded changes.
We do try to update the drawings to match reality (a revision known as “as-builts”) but perfection is very elusive IRL.
I was a chemical/electrical engineer by training and trade before moving into software engineering.
> Documentation is hard enough for software projects. Why would we expect documentation quality of construction to be better?
Because it is, by it's very nature and purpose. Practically all of the engineering deliverables are written documents. These documents held engineers liable to any defect. In engineering, documents matter the most.
Also, while you can make changes, you can’t really refactor a building or migrate it to a different service.
I think actually the problem would be placing the model accurately in a general indoor case.
No GPS, not that that is accurate enough anyway. There are other tricks to do it, restoring from a known point say.
When used in industrial environment, AR will use indoor positioning systems to make sure the model and real world are aligned.
Comments
Building services engineer here. I think AR could be useful for "x-ray" views of services above ceilings and in risers where you're trying to locate concealed equipment. When it's sitting in front of you in a plantroom you really just need to look at it or have a regular screen to tell you what's going on in terms of operations.
The big assumption is that the 3D model of equipment used by an AR app matches what's actually installed on site. This is rare in my experience - particularly over time as maintenance is performed in bits and pieces without documentation being updated. It's exceedingly rare to find an old building with accurate documentation of where everything is. I'd love an app that could compare a laser scan of installed plant with the 3D design model and automatically align them into an "as-built".
That's an interesting (and GOOD) idea! Like real life back-annotation!
It’s also impossible in the vast, vast majority of real-life situations.
As DeNiro said in the terrible Movie “Ronin” — “But the map…the map is not the territory.”
The 3D model will differ from the actual construction. From the very beginning and more and more over time. Documentation is hard enough for software projects. Why would we expect documentation quality of construction to be better?
I can see this being used for retrofit designing. Use a LiDAR point cloud to create a model of an existing open facility, all the exposed machinery and piping. Then suggest changes for future installation and have current staff walk through the plant checking the virtual future piping/etc.
We already do that without the 3D glasses though. Just 3D rendered to computer screen.
Not really. The point of construction is to turn the "3D model" into reality.
Some problems in the original planning can and will be spotted during construction but that just means plans are updated once conflicts are resolved.
There's even a occupation specialized in ensuring that what's being constructed complies with the "3D model": construction and building inspectors.
The mental models that are formed by software developers based on their personal experience within software development activities is not compatible with any day-to-day practices of any engineering field.
Because it is, by it's very nature and purpose. Practically all of the engineering deliverables are written documents. These documents held engineers liable to any defect. In engineering, documents matter the most.
Engineering is not software development, and software development is not engineering. Engineers need to put in writing stuff that works for decades, at their financial and even civil and criminal risk. You don't just hit the "build" button and retry if something failed. You put into writing exactly how something needs to be done in order to meet all requirements, and make sure that everything is done according to what was put in writing. If anything goes wrong, what's written down is the engineer's only defense.
Respectfully, I’ve done this specific job in the past for industrial/commercial building construction. It’s very, very difficult to get builders to build the thing to perfectly match the drawings. They often build what is convenient, or make absent-minded changes.
We do try to update the drawings to match reality (a revision known as “as-builts”) but perfection is very elusive IRL.
I was a chemical/electrical engineer by training and trade before moving into software engineering.
Also, while you can make changes, you can’t really refactor a building or migrate it to a different service.
I think actually the problem would be placing the model accurately in a general indoor case.
No GPS, not that that is accurate enough anyway. There are other tricks to do it, restoring from a known point say. When used in industrial environment, AR will use indoor positioning systems to make sure the model and real world are aligned.