The new Florida law passed last summer, SB-4D, is the legislative response to the Champlain tower disaster. Without getting too deep into the weeds of all the details, the new law states any building 3+ stories and 30+ years old (unless within 3 miles of salt water, then it’s 25+) must have an inspection done to certify the structural integrity of the building and assess useful life of the existing property (full law can be read here). As a consequence of this new legislation, building owners and associations are now scrambling to obtain these new certifications, as well as beefing up their reserves per the new guidance from SB-4D. Unfortunately, building inspection as it stands is not a cheap endeavor. One might assume paying a premium means comprehensive, high-quality datasets. Well, not always. In reality, the methods currently employed are some of the most dangerous, least cost effective, and produce practically unusable data. So, not only are building owners footing a big bill for a subpar inspection, engineers are being forced to make an impossible judgment of building integrity based on bad data. Add in the volume of buildings to be inspected in such a strained labor market, and you have a recipe for disaster. To get a better idea of the landscape, let’s look at the current methods being used by inspectors and assess safety, costs, time requirements, and data produced.
How is this currently done?
Today, engineers use four methods to find structural defects on the building exterior:
- rappelling down from the roof on ropes
- scaffolding
- bucket trucks
- expert with binoculars or long lens camera getting images from the ground.
When hearing about rappelling from roofs, walking on scaffolding, or hanging from a bucket truck, does your gut tell you these are safe? With the exception of the expert on the ground (which we will get to in a bit), all these methods have reported countless injuries and many deaths in 2022 alone. In fact, rappelling from the roof is considered so dangerous, insurance carriers no longer cover it. A slight wind is an extremely hazardous scenario for rappelling – picture a team swinging across a building when the afternoon sea breeze rolls in. Unfortunately, scaffolding isn’t much safer. As recently as January 2023, a scaffold collapsed in North Carolina leaving 3 people dead and many injured. (Alonso, 2023). Moreover, and if you want to be sickened, Google “bucket track fall” or “bucket truck failure” to see how often people are killed or injured operating them. As a rule of thumb, when a person has to be suspended in the air for any reason, there is an inherent risk, leaving only 1 traditional method that would be considered “safe” by any means – the person on the ground holding the binoculars.
Now that we’ve established most these methods aren’t exactly the safest, do any of these options seem cost effective? Well, rappelling from ropes is very expensive given the limited pool of people willing and qualified to handle the job. Scaffolding, forget it. Weeks of setup and takedown, not to mention permitting. Bucket trucks aren’t as exorbitant as the first two, but is still steep since you will need to factor in qualified operators, the truck itself, and insurance for at least a week or so – I could go on. What’s really shocking is the cost of the binocular inspection. The “expert” collecting data with binoculars will charge for a week-long inspection to only produce a handful of blurry pictures. I’ll let you in on a little trade secret, these inspectors are oftentimes taking the pictures from their phones hundreds, if not, thousands of feet away from the building. Can you say “high cost, low return”?
Right, but what about the data collected? Well, the people rappelling from ropes can only inspect what is directly below the rope. Contrary to popular belief, these people aren’t swinging around the building like Tarzan. Jokes aside, many areas will be missed unless you pay big bucks for the inspector to get a reasonable overlap which will raise your inspection timeline from weeks to months. Even then, the rappeler’s data will still be insufficient given the inconsistency of the different camera angles and potential lack of geo-tagged images. Scaffolding on the other hand, not only disrupts occupants of the building, it blocks much of the building façade you’re trying to inspect. For bucket trucks, the longest bucket truck readily available for renting can only serve a building topping out at 10 stories. If your building is over 10 stories, you’re out of luck. This leaves the worst data-collection option, the person on the ground with the binoculars or camera. At least with the rappelling, scaffolding, and bucket truck an inspector has a close look at the building façade to document anomalies and make decisions based on good judgment. The binoculars and cameras are limited to everything that can be seen from the ground. Finding a structural defect with these people is the professional equivalent of finding a needle in a haystack. In all seriousness, how would binoculars know, without a doubt, the severity of the defect viewed from an angle at a couple hundred feet away, or even fifty feet away, 20 feet away? What documentation can this person reasonably take from the ground? What measurements can this person truly derive? What if your view is obstructed from the ground, do they just skip that area? And this person is supposed to give an accurate depiction of the useful life left?
The old saying goes, “bad data will lead to bad decisions”. The bad data here can easily cost building owners millions of dollars.
New Methods
Now you have a basic understanding of where the industry is at currently, let’s understand why drones & AI will one day supersede all the current methods and be the sole data source for inspectors in the future. Going into the specs a bit, the camera mounted to our drone has a 48-megapixel full frame sensor that can produce a ground sample distance of 3mm per pixel resolution. I’ll pause here to explain why these specs are massively important for inspections. If you’ve ever seen something like a Google earth, Google maps, Apple maps, or other commercially available satellite imagery, you’re seeing a standard resolution between 2-5 meters (Maxar, n.d.). If you’ve ever attempted to zoom in on that satellite imagery, you’ll notice the overall image starts to get blurry and eventually unrecognizable. This is because a 2-5 meter resolution means that each pixel occupies 2-5 meters on the surface it is imaging. To tie that back in: the smaller the pixel resolution, the more detail can be derived from a single picture. So having 3mm resolution means your pixels occupy 3 millimeters of space on the surface being imaged (or in layman’s terms, it sees every little detail including hairline fractures on the building exterior).
Ok, the cameras are phenomenal, but the beautiful thing about utilizing modern technology is when you pair that camera with a drone, which can be pre-programmed to fly missions autonomously at a high image overlap. Plug in the distance from the building and the amount of image overlap you desire, and the flight software generates a flight plan covering 100% of the building exterior, including the roof. Just for fun, let’s say there is an area of the building that is really hard to capture. Remote pilots can manually fly the drone up to hard to capture area and rotate the camera gimbal into whichever direction is needed to capture it. 100% building exterior coverage at 3mm per pixel resolution is simply unheard of, even in the survey mapping industry. Add in the fact flight plans can be saved and stored for future inspections, and you’ve got a repeatable and replicable data collection method for your next milestone inspection to truly show wear and tear over time, something unseen by traditional methods.
Now here’s something to think about. Not only can drones capture 100% of the building exterior at 3mm pixel resolution, the camera payload can be swapped for a thermal camera. Thermal imagery detects water intrusion, the root cause of most structural integrity issues. Between covering the building exterior with high-definition photos revealing the tiniest of hairline fractures and thermal imagery showing anomalous water penetration, nothing is going to be overlooked or missed. Traditional methods simply cannot compete with modern technology.
Is New Technology Safe, Cost Effective, and Trustworthy?
Right of the bat, keeping an operator on the ground and out of air is a safety win. I’d like to point out that there have not been any reported deaths due to commercial drone operations (notice I used commercial, not military). Obviously, I wouldn’t go as far as saying this will never happen because crazier things have, but it’s worth noting commercial drones haven’t caused a single death in the last decade. Most of the reported drone-related injuries are lacerations caused by the propellers (Jaime, 2022). In those cases, the cause of the injury was a result of pilot error and ultimately the injuries sustained are to the pilot. Nowadays, there’s new technology that’s evolved to allow for prop guards to be installed on the drone to protect against these potential hazards.
But what happens if it falls? Luckily, these instances are quite rare, but complete failure is a possibility not to be overlooked. Two pre-flight actions need to be taken to combat against this worst-case scenario. The first being to completely cone off the area below the drone to the public so if the drone enters freefall, bystanders will not be directly below the falling drone to take the hit. Next, the best defense against any freefall scenarios is simply to install a parachute. A growing network of pilots (including us) are now equipping their units with quick-deploy parachutes for all operations near people. So, if a unit enters freefall, a parachute deploys within a second, slowing the descent speed of the drone and allowing the operator on the ground more time to warn any bystander nearby of the emergency.
Is drone data acquisition more cost effective than the traditional methods? Well, does water freeze at 32 degrees Fahrenheit? Does the earth revolve around the sun? The biggest driver of costs is how long the data acquisition takes. The longer the job goes, the more exponentially expensive the project gets. Drones can get high quality imagery of 100% of the building exterior in a single day in most cases. Can any of the traditional methods boast the same thing? Not even close. Traditional methods are usually taking anywhere from several weeks to months. And I haven’t even mentioned the cost savings in the office work (more about this below).
Can this data be trusted? Fortunately, drone collection has redundancies built in. The number of photographs overlapping each other is a check against a false negative. Meaning, if one photo shows something suspicious, another two or more photos can be brought in to verify if there is a real issue or not. Also, every photo is geotagged and post-processed against a static base station, which gives the processed data a geospatially-accurate location. Having a geospatially-accurate location means photos will not be placed in the wrong area (photos taken on the south side of a building will not place with photos from the north side of the building, etc.).
Here’s the bigger question: What would you have more trust in: a geospatially correct dataset showing 100% of the building exterior with imagery at 3mm pixel resolution, or what you can see from the ground a hundred feet away on the inspector’s iPhone 6? There is no comparison. One dataset offers you complete coverage, and the rest cannot.
Leveraging AI Software
I like to think about building inspections as a marathon race. The traditional methods are your seasoned runners who all finish around 2.5 hours, which is a great pace albeit a bit grueling. The drone and software technology is the Porche 911 who enters the race and effortlessly finishes in 16 minutes. As you recall, the drone can capture 100% of the building exterior in very fine detail. Well, once the acquisition is completed, all the photos are then uploaded into SFM (structure for motion) software where beautiful and geospatially accurate 3D models are completed. These models are important for many reasons, one of which is being able to retroactively get a view of the building at a specific moment of time and mark any significant variations on future inspections. Imagine if there were historical 3D models of the Champlain Tower completed leading up to 2021. An engineer could have layered the historical datasets and seen the deepening cracks, increased water intrusion, and crumbling concrete and made a sound determination of evacuation due to safety concerns. Instead, engineers relied and continue to rely on cartoonish drawings created from bad data to decide a rate of decay. These days, not having a digital library of a historical building data is borderline negligent.
AI also significantly trims the amount of time needed in the field and in the office. Instead of pouring through endless photos attempting to annotate every crack, AI completes this task automatically, and with measurements. This means a defect is positioned, annotated, and given a severity score on a digital model without the need for human intervention. How would an inspector measure the size of a crack nowadays? Well, they whip out their handy dandy ruler and start taking measurements. But what if there are hundreds if not thousands of cracks? Let’s just say the inspector will not be measuring every single one of them. AI can not only measure every crack, but if the crack is wide enough, the AI can also perform a depth measurement of the crack.
This process makes engineers lives less cumbersome. Engineer approval is made easy by simply logging in to view and verify the defects, saving their precious (and expensive) time for other tasks. Outside of the incredible time savings and cost-effectiveness this technology provides, engineers no longer need to be shuttled to the field for benign tasks (like verifying a crack with their naked eye) when they can rely on quality datasets.
Technology is the Future
It is only a matter of time before state agencies require drones as part of a building inspection for documenting conditions. The traditional methods are simply too dangerous or produce such bad data that a change is inevitable. Plus, it doesn’t make monetary sense for building owners or building inspectors to keep propping up these old practices. We shouldn’t allow SB-4D to become another boon for the old guard. Let’s leverage today’s technology and give engineers and building owners what they want: complete, high-quality data sets so important decisions on a building’s structural integrity can be made with confidence.
Dylan Robbins
Founder, 1-Day Inspections
Works Cited
Alonso, M. (2023, January 2). 3 dead and 2 hurt after a scaffolding collapse in Charlotte. Retrieved from Cnn.com: https://www.cnn.com/2023/01/02/us/charlotte-scaffolding-collapse/index.html
Industries, B. t., & Bradley, a. S. (n.d.). CS for SB 154. Retrieved from Flsenate.gov: https://flsenate.gov/Session/Bill/2023/154/BillText/c1/PDF
Jaime. (2022, December 20). US Drone Statistics 2023. Retrieved from https://skykam.co.uk/drone-statistics/
Maxar. (n.d.). Clarity and Confidence. Retrieved from Maxar: https://explore.maxar.com/Imagery-Leadership-Spatial-Resolution#:~:text=Most%20commercial%20imagery%20falls%20between,50%20and%2030%20centimeter%20resolution.
Tolan, C. (2021, June 29). Condo owners in Surfside building were facing assessments for $15 million worth of repairs. Retrieved from CNN: https://www.cnn.com/2021/06/28/us/surfside-condo-owners-assessments-invs/index.html