Extracting Consumable Construction Intelligence from Reality Capture Data

CHIDAMBARAM SOMU: So before I start-- a couple of housekeeping items. When I get excited, I tend to speak fast. So if you notice I'm speaking fast, slow me down. And this presentation is not about any product or any software. This is more a process and workflow related presentation.

So just a quick hand-raise, how many of you guys are from the construction industry-- project engineers, managers, superintendents? Good. Good, quite a bit of people. So how many of you guys from the vendor side, like drones, laser scanning, reality, capture, any kind of vendor side? Cool.

So the construction population is majority. That's good. But that was the expectation for the class, because we're going to talk more about process and workflows. Cool.

So my name is Chidam. I'm with the DPR Construction. I'm based out of our Phoenix office. I've been with DPR for about seven years.

And my area of interest or expertise, I would say, started as a BIM engineer, BIM manager. And then I was managing the BIM implementation for the region. And I had a little pet project.

DPR is more entrepreneurial, which is the reality capture. Which I ended up starting the reality capture business unit for DPR. And it's been going pretty good.

So I just wanted to share the lessons learned in the past few years throughout the journey of how we have encountered reality capture processes and workflow. So little bit about DPR Construction-- do you guys know about DPR construction? No?

So DPR Construction, we have spread across the nation. We also have offices international. Our headquarters is at Redwood City. And last year, our annual volume is like $4 billion, and then Fortune 100 companies to work for in the top 10 companies, and then rated as number 20 in ENR magazine.

So one of the main reasons we got into the highly technical reality capture stuff is because of the core markets we deal with. So we deal with the core markets which are highly technical-- data centers, pharmaceutical, biotech, health care, higher education, which are all highly technical building. So we wanted to invest in technology on workflow, so that we can reduce the rework on waste. Because one of the biggest profit eater in construction industry is waste and rework.

So how we can use reality capture to reduce the waste is the whole theme behind the presentation. So a quick intro about reality capture-- so as I said earlier, this is not going to be a presentation about a software product. This is more on workflow.

So reality capture is a process of capturing as-it-is condition. It doesn't matter what you capture with. Let it be an iPhone camera or a high-end laser scanning device. So reality capture is a process of capturing as-it-is conditions.

So we put reality capture into three buckets. Right now, we are dealing with three different approaches. The one is where the lighter, which is very conventional. You use the laser scanner, capture the as-it-is condition.

And the second bucket which has been getting popular in recent days is drone technology. So we use drones to capture the as-it-is conditions. And also, along with 360 cameras right now, there is a lot of products you might have seen in the convention, the expo.

They're coming out with a lot of 360 photographs and products, which are a little bit more intelligent. I'll talk about it down the slides. But that is another mode of reality capture.

And we also add another bucket, which is data sensing and visualization. So you capture all this reality as-it-is conditions. And there needs to be a visualization component to it. So there are new technologies coming up with AR, VR, MR, so you can actually interface with the data you capture. So that is more of a data sensing and visualization bucket.

So these are all the three buckets we are dealing with right now in reality capture. So the reality capture, as I said, the data can be from any single source or combination of sources. But if you look at the products available in the industry, is it catered to one product, maybe a drone? Or some products are located to one instrument, which is drones or laser scanners.

But reality capture, the data can come from a 360 camera, or your iPhone photo, or from a total station, or from a drone. Or you might have heard about the backpack scanners, which is also getting so popular right now. You put on a backpack.

It has a lighter sensor. You just walk around the room. It captures the as-it-is data. So it is getting popular. And with the headsets, they're also coming up with these [INAUDIBLE] and other helmets which capture the photographs.

So the source can be from a combination of sources, just an FYI. Even if you capture an as-it-is condition through a total station, that is also reality capture data. So it can be a combination of multiple things.

So just a quick intro about LiDAR, have you guys used laser scanner on job sites, just a quick hand-raise? OK. All of you guys have used laser scanner. Have you guys used drone on job sites? OK. Good.

So LiDAR, I don't have to go into details about how it works. LiDAR is based on laser beams. It shoots a laser beam. Based on the reflection and return, it just calculates the position of the object. So that's basically LiDAR.

And photogrammetry-- you fly a drone. You capture multiple images. And then you tie all the images together. That images will be converted to a 3D point cloud or a contour drawing or an [INAUDIBLE] image. It could be anything.

So here is a snapshot of a stitched photo which we call [INAUDIBLE] photo. And then that is the output, which is the point cloud output. So this is where it gets interesting.

Because we got all these products. And we got all these different softwares that has been marketed for reality capture. There is ton of softwares out in the industry.

So when we came into reality capture, that was the biggest hurdle. Because we had to go and evaluate each and every product. Because there is no one product fits our needs. Because there is no one product available in the industry that fits all of our needs.

So there is different use cases. So this whole thing we are going to talk about is more based on use cases. So what did we use for a specific use case and construction-- so, again, coming back to reducing rework and reducing waste.

So there is pre-construction use cases. And we're going to see post-construction use cases. And then we're going to see construction use cases.

So I hope you guys are all familiar with pre-construction use cases, which is very typical. It's been in the industry for quite a bit of time, which is capturing the exact builds. So pre-construction use case is nothing but, if it's renovation building, you go in and capture the reality data using a laser scanner or using a drone for pre-construction purposes, like estimation, documentation, design, and coordination.

It's getting pretty popular among architects and design team. Previously, they had to invest a lot of money. It's a heavy labor intensive work.

They send guys, tape measurements. And then they take a total station, shoot points, and gather the data, which is very inaccurate. And it's not reliable.

So laser scanner, we can get to a 1/8 of an inch accuracy. That is more reliable. And right now, it is more popular in pre-construction capturing the as-it-is conditions.

So some of the use cases is documentation. When we go into documentation, if you're taking over a project or we are going into renovation project, first thing we need to do is capture what it is right now, so that we can build on it. And then the next one is design coordination.

So we capture the data. So when we go into renovation or TI fit out. They're going to add a whole lot of items to the existing condition. So we need to make sure there is no clash, and it's clash-free.

And then there is also a supplemental survey, which is getting very popular with drone right now-- site selection, contour mapping, cut/fill analysis. So site selection-- you have a big site, which is like 20 acres, 30 acres. So it's really hard to get a laser scanner out there and collect the as-it-is conditions of the site.

So that is when we use drones to collect the data and use for site selection, or contour mapping, cut and fill. It's super quick. So you just fly a drone. You get the data, and then you have the as-it-is conditions.

So we call this exact building in pre-construction. There is a difference between as-builds and exact-builds. Because as-builds are never accurate. I mean, you guys are from construction industry. You know as-builds are never accurate.

So when you start with an as-build, definitely you can encounter an error. I can assure you that. So exact-builds are nothing but capturing the as-it-is conditions with a very super high accuracy.

Here are some snapshots of the project. I'm not going to go into the detail, because I'm under the assumption, I'm making the assumption, that you guys all know about collecting as-it-is conditions. There are some interesting things. We'll look at it.

So this is the difference between exact-builds and as-builds. So as-builds, it's all hand-drawn drawings. End of the project, you'll see all these subcontractors will just draw everything and hand sketch everything and give it to us.

Exact-builds are more accurate. We know exactly where things are with relation to the space. So here is a quick snapshot. This is the laser scan data we collected for a project.

And then this is the model we created. So this is a TI fit out. Definitely, you would like to know where the utilities are before you go and put in your walls. So I'll just from this animation.

You can see how accurately it's capturing the as-builds conditions. Especially for TI project, if you go and add in everything, this is super helpful. And this is another snapshot of a central plant where we are tying into an existing building.

So it's an expansion project that is a huge $80 billion building coming up. And it's tying to the existing center plant, which is nearby. So we had to go and capture the as-it-is conditions of the central plant.

Here is a quick snapshot. And then what we do is we convert the point cloud into 3D models. So I wanted to give a quick intro of converting point cloud into 3D models.

When we capture the point cloud, that data cannot be consumed directly. So we had to either convert that to a 3D model, or we have to pull dimensions. Because when we run a coordination, when we create a Revit model, point cloud, as it says, it's a cloud. You cannot clash it with an object.

So it is when you put your hand, it's going to go inside. So it's a cloud. It's billions of points. So we take those points. We convert into a solid object, so that we can use it in a Revit model or clash with it a Revit model or a [INAUDIBLE] model, so that it can be helpful for the coordination. So that is what we have done here.

So site selection and survey supplement-- I think I missed my drone symbol here. So this is a drone project. So I'm going to just mix and match many use cases-- pre-construction, laser scanning, drone. This is a drone project.

And we had about a 20 acre site where we need to go and capture the contour. This is not for the actual use of the data. So this is for site selection.

Think about 10 years back. If we wanted to capture the existing conditions for a site selection 20 acres of the land, you would spend a month going and capturing all the data. So this project was done in three days.

So we flew a drone. And we captured the as-it-is conditions. And then we created a contour map and a heat map. And then we gave them, in three days, which [INAUDIBLE] scheduled about, like, a month. So initially, when they planned, they had scheduled to do a site survey so that they can select a space for the building.

And another interesting part-- I mean, I'm from Arizona. This project is in Coolidge. Coolidge, the surface is so undulated. So you've got to capture high accurate data. So we were able to do that with drone.

As the technology advances, we can also capture more accurate data. Because right now, the technology-- in an uphill trend. So there are so many new drones coming up, so many new different softwares coming up. So it is going good for construction industry.

So this is another site selection service supplement. This is a contour map that is created along with the heat map. So don't get me wrong.

When we go into a project, we always trust the information that is given to us from a construction industry from a construction standpoint. If we get a civil drawing, we go and say, OK, this is the civil drawing. This is going to represent the site we're going to work with. But that is not true all the time.

When we make cut/fill analysis, all our estimation is based on the civil drawing given to us. But when we go and actually do a site scan on the job site, it is totally different. In terms of how much you need to cut and how much you need to fill, it's totally different.

So in pre-construction for estimation, this is a great tool. Because we can validate the data given to us before we take that risk going to the construction and actually doing some work. So this is a great tool for estimation and pre-construction.

And this is another interesting topic. This is a 360 image. So right now, there are different products which can extract intelligence. Previously, 360 images are just images. There is no intelligence to it.

Right now, were you guys at the Smartvid.io talk yesterday? So that is extracting intelligence from the photo. So the industry is moving towards more data extraction, intelligent extraction.

Previously, it's all dumb photos. Now, we can pull dimensions from the photos, because the technology is advancing. And also, they are doing some machine learning to identify safety hazards and quality issues. I saw that Smartvid presentation. It was awesome.

So we can extract intelligence from the photos. This is a quick snapshot. This is a project in Phoenix. It's called Banner Medical Center, Phoenix. And we had to build a 17-story patient tower on top of an existing office space.

So that is a huge utility tunnel running under side. And we wanted to capture the as-it-is conditions, which design team can go and use it. We did a laser scanning on this job. But it's not directly consumable by the design team.

So we are to give them something which can be consumed by them. They can pull dimensions. They can work off of. They can do coordination.

So we had to take a 360 photo along that. You can see the 360 photos are actually machinable. It's browser-based, cloud-based. You go and measure the dimensions from the 360 photos. This is pretty cool.

So this is where it gets interesting. So a lot of the laser scanning we have done, a lot of the drones we have done, is more focused on pre-construction. But construction is where a lot of waste is being created, a lot of rework is happening.

So there is a statistic I'm going to pull up-- how much we are spending in rework. If we make a mistake, we'll go and fix it. But the subcontractor or anybody who have done a mistake is going to-- he need somehow push that cost into another change order. So the ultimate goal is to actually reduce the rework which saves money for the project.

There is a little investment to it to actually do the work, but it's worth it. I'm going to talk about the statistics a little bit later. So there are couple of use cases I wanted to address today.

This is all some of the use cases we are working on right now in construction realm using reality capture. The first one is verified field installed check as-builds, and prefabrication. As prefabrication is getting popular and popular, we need highly accurate as-it-is conditions so that the pre-fabricated parts can fit in.

So we are working on some exciting projects where the drywall is being pre-fabricated. The exterior skin is being pre-fabricated. You just go and host it.

But before we host it, we need to make sure the edge of slabs are at the right location, the embeds and the edge of slabs are in the right location, so that we can go and do the pre-fabricated skin on the project. So this is super helpful during that phase. And then floor flatness, floor levelness, ASTM 1155 reports-- there is conventional dipstick method of gathering ASTM 1155 reports. I'm going to show you some exciting workflows that we have been using to capture that report.

Floor flatness, floor levelness, quality control-- that's the huge thing. That's a big savings, construction savings area. And then I'm going to talk a little bit about camber analysis, site analysis, safety, and progress tracking. And that progress tracking is kind of evolving right now. So there is a lot of universities working on different researches to automatically track the progress.

Pre-pour QA/QC, this is one of the interesting topic we ventured into a couple of years ago. So before the concrete pour, you might have seen all these leaves and blockouts. Embeds, edge of slab, goes in place, as per the design drawings.

And then after we pour the concrete, we feel like, oh my god, this sleeve is completely out of place. Now we need close it and cut it at some location. Oh my god, this edge of slab is completely out of place.

So what we did is before the pour we went in and captured the as-it-is conditions. We captured everything in terms of where the blocks are located, where the openings and sleeves are located, where the edge of slabs are located. And then we have the design drawing, the model, the intended drawing in Revit, and also in CAD.

So what we did is we took the laser scan data that we created. And then we overlaid into the actual opening drawing. And then you know what? We found out a hell a lot of issues that has been fixed before even the concrete is dry.

So you can see there is a discrepancy in terms of the block out. Think about the amount of time we spend in BIM coordination coordinating all these mechanical, electrical, plumbing, all these utilities, stub ups, and all these things. And finally, everything comes down to if all these things be coordinated, is it right in place, or it's somewhere else?

So the problem is they are pulling tapes, and they're pulling chains. There is always a human error component to it. So we were able to identify that.

It's more a cultural thing. When we actually went and did that, the guys didn't trust it. The guys in the field didn't trust it. They had to go and see. Is it really worth it to actually see it?

And then what we did is we actually went into the job site and pulled the tape. It's accurate. What we show is accurate. The sleeves are off.

AUDIENCE: Did you have [INAUDIBLE] benchmark?

CHIDAMBARAM SOMU: Yes.

AUDIENCE: [INAUDIBLE]

CHIDAMBARAM SOMU: Yes. So that's a very good question. Because the benchmarks, when they come in and establish these benchmarks, this is a huge topic I can talk for hours. This is one of the problems that we are dealing with day in, day out.

So what we did to resolve this issue, this is a quick tip not related to reality capture. What we did is when we started this coordination, all the mechanical, electrical, plumbing subcontractors and the GCs established a common control, maybe in the corridor, or maybe in the place which is not going to change, and then said, this is the controls you're going to use throughout the projects. And the survey is going to lay out what we ask for. Not-- he's going to do some lay out and tell us where it is.

So we have established an offset from the grid. Say, this is where you're going to lay out your points. So that offsets were communicated to us. When we went into and laser scan, it was easy enough for us to actually orient it to the project coordinates, so we can actually verify the as-build.

So that is the huge thing. We learned some hard lessons throughout the course of the implementation. But now we have a standard process, what needs to be done in terms of lay out and benchmarking. So everybody is following that.

AUDIENCE: So what do you do on your benchmark [INAUDIBLE] for the x,y coordinates pretty good [INAUDIBLE] building settles [INAUDIBLE] over time [INAUDIBLE] building settle [INAUDIBLE] thing. So do you have a workflow for every measure in your z coordinate?

CHIDAMBARAM SOMU: I mean, there are some things that we are doing. That's more of a best practice. We just triangulate. Whenever we come in, the surveyists come in every week and triangulate and do a level loop around the building to make sure the z is set at the right elevation.

Because it's more of a best practice rather than a workflow. And we said every week, you're going to come in and run a level loop and make sure all the z coordinates are tightly set. So that is one of the things where there is no answer, but there are best practices to avoid that.

But here, we are dealing with more of x and y. And there is always a problem with a human error-- I mean, human implementation-- where they pull the tapes, they miss a couple of inches here and there. We were able to fix that.

And also, laser scan becomes a second set of eyes to make sure that the benchmark they have laid out is right. So if that is not right, it's obviously going to show up in the laser scanner. Our software is going to throw up some error saying the benchmarks are wrong.

This is a huge thing for us. And this is another one. We kind of advanced. Previously, you might have seen just pictures from that CAD. Now, we've advanced making reports where we have this 20 page report, which is generated after we go and do the pre-pour QA/QC, how much it's deviated.

And also, we started giving some [INAUDIBLE] for the right location. So we started indicating that, hey, you did it right. 80% is right. 90% is right. It becomes like an encouragement for the field guys to actually get 100%. So something like a reward-- say, hey you got all 100% of your sleeves right.

This we kind of evolved and started developing reports. Then we encountered a problem of communication. Because these reports are 20 page reports. And they won't be able to look at each of these reports and go and fix it in the field.

So we were actually communicating with the supers right then and there, so it can be fixed. There is different strategies we had to use for the communication. But this is all workflow.

So as I said, if you see that symbol, that's the symbol I missed. So if you see that symbol, it's a drone project. This is a very similar thing, but we are actually doing it with drone.

Have you guys seen any the as-build underground utilities are right when you get it? Right? So this is one thing that, I mean, even in construction industry when you go and put an underground utility, after a month back, you come and part hole and find out where are the utilities. That's how the industry is right now, because we don't even know where we put in the utilities.

So this is a great workflow. So after we excavated, we put in a first level of underground utilities, we fly a drone. And then it captures all the as-it-is conditions. We backfill it. And then you put another next layer of utilities. You fly a drone, and then you as-build it.

So end of the day, after the ground is scoured, you see this whole underground stack of utilities where you can overlay your civil drawings. Any time, you can tell within a couple of inches. A couple of inches for underground utilities is super good.

So within a couple of inches, you can see where this waterline is exactly in place when we actually did the construction. So this is a great as-build handover for the owners. Owners are going to love it, because they don't get a true as-it-is condition from the underground.

But this is more related to QA/QC. So you can see all the conduit stub ups coming up of the ground. That is the actual field location. But the intended location, the model is different.

So we can fly a drone and the overlay the point cloud we collected on, also, the model and find out if that is a bust or not. So the technology is advancing. Now, you can get to an accuracy of couple of inches, close to a couple inches. And also, a lot of products, like, Autodesk ReCap or you get into different products, they have incorporated the global coordinate system where you can hook it up with global coordinates.

And this is the statistical analysis I was talking about. After the project we completed, after a couple of projects we completed, we wanted to prove it to the project teams and owners that we actually saved money. Because unless you show a monetary value to it, nobody is going to just buy into the workflow if that workflow is pretty cool, right?

So what we did is I worked a couple of months working on a paper. This was submitted to American Society of Civil Engineers. This is a paper on rework savings in QA/QC using laser scan.

So what we did is we collected all the errors, which we might have encountered-- so the errors which we fixed it in the field using laser scan-- and associated a monetary value to it and said, if we wouldn't have used laser scan, we wouldn't have encountered this much amount of rework. So that is the theory behind it. So we said cost of laser scan to do per square feet is $0.10.

But the potential cost of field rework that we saved using lasers can, all the errors you have seen in the document, we fix it. We put in a monetary value to it and found out it came down to $0.85. If you have 1 million square feet of a project, you can potentially save-- it's all potentially you can save-- $850,000, close to $1 million on a project, which is a huge cost savings for a project.

And think about the schedule savings, to write an [INAUDIBLE], to move and embed, or to write an [INAUDIBLE], to move a sleeve, or anything like that. So that is an administrative time. There is a cost to the project engineer's time, civil engineer's, structural engineer's time to respond to the data [INAUDIBLE].

There is a whole another different story. There is a schedule component to it. And there is a cost component to it. So this is dealing more with a cost component of how much potential savings that we can make.

And floor flatness, floor levelness, this is also evolving right now. I'm going to start with quick snapshots. This is also with laser scanner. So we captured the as-it-is condition.

And then we generated this cool 3D heat map, which shows the highs and lows. But this is not acceptable by The Concrete Society, because we need to submit an ASTM 1155 report, that's a floor flatness, floor levelness report, to comply with the spec, right? So this is pretty cool.

Now, it's more like a post-mortem. After you pour the concrete, you know where the highs and lows. It's easy to go and fix it. But this is not acceptable by the spec. It's not compliant with the specs, because it's not producing an ASTM 1155 report.

And then we got into a software that can actually produce ASTM report, which is a lot more accurate. Think about dipstick. If I make a run right here, the floor flatness is perfect. Because that run does not have any undulations.

But if I make a run right here and it's going to throw off and show so many errors with floor flatness, so it is not a true method to actually capture ASTM 1155 reports. I've seen guys, like, do the dipstick and capture it and look at, OK, there is an error. Let me move two steps, and then he'll run the dipstick and find out, OK, this looks good. Let me comply with that.

So this is more accurate method. So we were able to produce ASTM 1155 reports. This is our own report that we hand over to the superintendents and subcontractors, so they can go and fix it. And here is a sample ASTM 1155 report, which is generated from the laser scanner which is compliance with the spec and which is acceptable by ASE.

So you can see the runs. This is laser scan data captured on the floor. And you can see we have made a couple of runs. And then automatically, it produces a report if it's a pass or a fail. So this ASTM 1155 report is fully generated from the laser scan point cloud data. Any questions on this?

AUDIENCE: So we've been doing this as well. [INAUDIBLE] drywall stuff. And we recently got [INAUDIBLE] scanners [INAUDIBLE]. We recently started [INAUDIBLE] floors. We have issues with [INAUDIBLE] like that I think could be a big issue for us. So I guess my question is what do these things actually fix? I mean-- yeah.

CHIDAMBARAM SOMU: Yeah. Very good question, very good question.

AUDIENCE: Did your concretes have actual [INAUDIBLE] in it?

CHIDAMBARAM SOMU: So we are using a little bit different workflow. I'm going to talk about it, what it is. Because it's a post-mortem. So if we need to grind the concrete, you have to go and grind the concrete.

But grinding concrete is going to cost more than actually fixing the drywall stuff. It's a project team's decision. But we are also experimenting a couple of new workflows. It's called wet concrete scanning.

So previously, it takes about a couple of days to actually start to finish. You go and scan it. This is what I'm talking about.

Five years ago, you go and scan it. You collect the data. And then you post-process the data. And then you produce all these beautiful reports by the time the concrete is dry, and you cannot do anything. It's post-mortem.

And then now the technology is advanced. So we scan in the morning. In the evening, we give them report. But still, it's not good.

What we are trying, recently, it is the experimentation we are doing. It's called wet concrete scanning. So when the concrete is wet, you go and scan it. So the softwares and ASTM reports are not going to help you.

Sorry-- jumped ahead. So we are using this heat map report. So while they are pouring the concrete, when they are doing the initial step, we are actually capturing the scanned data. And producing this heat map is going to take you, like, minutes.

So we produce the heat maps. We don't produce any reports. We show them in the computer, because we exactly know where the location is. We'll say, hey, go and fix that. Just smoothing it. That's high, that's low.

So this is wet concrete scanning where you can actually fix before the concrete is dry, so we are not dealing with post-mortem reports. Does it make sense? This is a new workflow that we are experimenting, because the technology's advanced.

This is what I'm talking about, extracting intelligence that can help us to save rework. When the concrete is dry, there is no use in doing that.

AUDIENCE: Are you using Rithm to do those ASTM reports?

CHIDAMBARAM SOMU: Yeah.

AUDIENCE: OK.

CHIDAMBARAM SOMU: But this one is from a different software. The ASTM reports is from Rithm.

AUDIENCE: OK.

CHIDAMBARAM SOMU: We are using RealWorks for this one. So this is the actual representation of the floor, ASTM 1155 reports. It's not the true representation of the floor flatness. So we doing wet concrete scanning.

This is an experimentation which we are working on it. We haven't come out to a point where we can go and release the workflows. But we are working on it.

Here's a quick statistics after I talked about all these pre-pour QA/QC, floor flatness, drawn QA/QC. So this is the report which s generated by Pharaoh. 5% to 12% project cost is wasted on rework and schedule delays. This is the exact waste I was talking about.

By doing pre-pour concrete QA/QC, you can save $0.85 per square feet. On 100,000 square feet, it's $85,000. So it's driving down the rework cost, base cost, until we actually-- this is hard to showcase. Because this is all assumptions, right?

We're going to assume that the things have not happened. But if it's happened, it would have cost this much. So it's all basically assumptions. And then 5% to 7% rework total installed cost of a Brownfield project and driving rework contingencies down to 2%-- so this rework contingency is more related to QA/QC, pre-pour laser scan, pre-construction laser scan, those kind of things.

And we are also working on some recent advancements with camber analysis. Have you guys used laser scanner for camber analysis? OK. Good.

So camber analysis, this is also Rithm software. This is not about the software. I'm going to show you another workflow.

There is one part, which is required to compliance with the spec. There is another part, which is going to actually help to solve the problem. We are also trying to come up with reports that can compliance with the spec. But actually, we are more looking into results that can help us to resolve the issue.

So this is good for reports. But we wanted to find out something which can actually help us to solve the issue. So here is a camber analysis report.

We laser scanned it. And then we were able to actually put points and find out the camber. On a conventional workflow, you take your tape measure. You just put it how much it is from the floor. You find out how much camber goes out.

But here is an interesting approach. This is just tweaking the workflow a little bit different. We made a cut plane. Say, where there is a high camber, it's going to highlight you with red which means there is a high camber.

And then you'll get a heat map saying that where the camera is high and where the camera is low, but it's not actually producing the actual cambers. But this is going to help us in terms of fixing issues. So we are looking into workflows that can actually help us to fix issues. So here is a quick snapshot of that. You can see the blues are low cambers and the reds are high cambers.

And the good thing about it is one thing I want to mention about camber analysis or pre-pour or post-pour QA/QC is you don't have to go to the job site every time to do this. It's one-time data capture. If you do pre-pour QA/QC, by the time when you do pre-pour QA/QC, the steel is already erected.

You might need to go and find out after the camber is poured. So when you do one-time data capture, you can use the same data to do pre-pour QA/QC, floor flatness, post-pour QA/QC, and also camber analysis. So that is the beauty of it.

We don't have to spend labors every time to collect the data. The data collection is one time. And you're using the data to do multiple workflows, which can help us with a project. So that is the beauty of laser scanning or drawn or capturing this data.

And here is a site selection cut/fill analysis. I already talked about it. And we always go with an assumption, the civil drawings given to us is pretty good.

But we try to break that assumption. And then we went in and did an entire drone flight. Since this is all cheap, it's not going to take a lot of time. You can do it on the fly.

And we were able to come up with a cut/fill map which can tell us where to fill it and where to cut it. And here's a quick stockpile analysis for excavation. A lot of drone softwares are doing it right now.

But this is more drone specific, right? If you buy a drone software right now, it does cut/fill analysis. It does a site analysis, contour analysis. But this I'm going to talk about a little bit. I don't want to open it up right now.

So any product you take it's going to do stock pile analysis. That's pretty common use of drone. And this is another thing that we are trying to do with the safety right now.

Because the drone picture has taken every day or every other day some projects and weeks. But every day, we try to fly the drone in the morning which tells us where the materials are stored and how the site fencing looks, how the traffic flow looks. Everything is captured in the drone data. We are using that for safety and communication.

Because we are tired of using Google Maps, which is like probably a year old. Especially with a job site, you don't have an accurate data to do site logistics. So we are flying drones every day.

So what is the use of flying the drone every day? That is another thing. As I said, the data is captured once. But the data can be used for a different purpose.

When we capture the drone every day, we can actually simulate the progress of the job. That is the research that a lot of universities are doing. So you can simulate the project. But once it's covered by the building, that is where the industry is lagging right now.

There is no one which takes different data where we can look at from a top view, find out the construction progress. Once the building is up, you cannot track being interior progress, or you cannot track the exterior progress. So that's one area that a lot of research has been happening.

But this is what I'm telling about. It's not actually taking the fall. It's actually extracting the intelligence from the photos to find out if there is a safety hazard or if there is a potential site logistics problem.

And this is more of a progress walk-through. I'm sure you guys might have used some kind of a 360 camera with progress photos. This is a product which was called StructionSite, where you can actually drop in pins, and then you gather the 360 photos.

Previously, what happens in the industry is you take photos, and then you come back. And then you open up a floorplan. You tie in all these photos.

Right now, as I said, everything is advancing. So you take your iPhone. The floor map is loaded. You just walk in. Then you can also mount your 360 camera on your hard hat. You just walk in with your floorplan.

Just click. It drops a pin and takes a 360 photo. You walk to the next part, drop a pin. This is a great way to communicate to the owners. Do you have any--

AUDIENCE: [INAUDIBLE]

CHIDAMBARAM SOMU: Previously, it was Bluebeam. I mean, Bluebeam workflow is a little bit complicated in terms of you have to take photos, and then come back and tie the photos. This is a StructionSite, where you actually walk in with the floorplan on your iPhone. Just tap on it. Tap on it. It just captures the 360 photo and puts it on the floorplan map.

This is great for documentation. As I said, this is more of a construction workflow. It can have the ASE meetings and stuff with the owners.

So talking a little bit about post-construction-- so as I said earlier, there is three buckets we are dealing with, three different types of use cases. One is pre-construction and construction and post-construction. Post-construction use cases are evolving.

There is so much amount of potential right now with post-construction use cases. It's evolving. One of the quick and dirty things we can do is documentation. Because we are collecting all this data, as I said, data collected once.

We can use it for documentation, handing over to the owner. There is so much amount of valuable information. I'm going to show you this snapshot.

So we've been getting requests from the owners that when there is a problem, when there is a leak, nobody knows what's behind the wall, right? There is plumbing lines running behind the wall. There is pneumatic tubes running behind the walls. There is no information on where those pneumatic tubes are, where the plumbing lines are.

So we were able to actually capture the laser scan. This laser scan was captured way earlier. But we were able to overlay the laser scan with our drywall model and tell where exactly all the plumbing pipes are, where the junction boxes is, and where the pneumatic tubes are.

So this is a great tool for as-build documentation. When the owner goes and fixes a leak or when they run into any situation, emergency situation in future post-construction, they can refer back to this. Repeating.

Also, this is one area which is getting super, super fast. It's post-construction facilities management. So after we turn over the data, as I said, the as-builds are hand-drawn. But there is no true representation of the as-builds which we call it exact-builds.

So if the owner goes and fixes an issue, if he has a VAV box above a ceiling, that VAV box may have moved in the field, but it's not capturing in the as-it-is conditions. So I'm going to show you a quick example. We wanted to practice what we preach.

So before we went in and published the workflow, we went in and tested it for our own office. This is the office I work out of. This is the Phoenix office.

And we didn't do any good as-build documentation after construction. So what I did is after the construction was done, five years after the office is up and running, I went in and captured all the laser scan data. So lasers can data is good in terms of coordination. But there is no intelligence to it.

So what we did is we built intelligence using BIM 360 Field and Glue. So you model everything. You push intelligence into it.

So extracting intelligence from submittals, extracting intelligence from operation maintenance manuals, extracting the maintenance schedule, everything, and then push it into the model-- it can be done way simpler during the construction process as well. But since it's completed facility, we were able to capture everything and collect all the data, push it back to the model. So that model can be used in post-construction.