Your Foundation Guy Checked 2,000 Square Feet With a Ruler. The Laser Finished Before He Uncapped the Chalk.

A concrete finisher named Dave is checking a slab in Abilene. He's got a 10-foot aluminum straightedge, a box of chalk, and twelve years of experience telling him what flat feels like through his boots. He sets the straightedge down, peers underneath it, moves it six inches, peers again. Repeat. For hours. His slab is 2,100 square feet, and when he's done, maybe three and a half hours later if nobody interrupts him, he'll have checked roughly 8% of it because that's all the straightedge method can realistically cover in a day, which means the northeast corner where the pump truck operator rushed the last few yards before quitting time will remain a complete mystery until someone tries to install tile twelve weeks later and discovers a surface rocking like a barstool on three legs.

Meanwhile, inside a Department of Energy lab in Oak Ridge, Tennessee, a group of researchers just did the same job. Forty-seven seconds.

What FLAT Actually Does

Government researchers have the branding instincts their acronyms suggest. FLAT stands for Flat and Level Analysis Tool. It was developed at Oak Ridge National Laboratory's Building Technologies Research and Integration Center by Nolan Hayes, Bryan Maldonado, Mengjia Tang, and Diana Hun, published in June 2025, and presented at the International Symposium on Automation and Robotics in Construction in Lille, France.

Beautifully direct concept: set up a terrestrial laser scanner on the slab. It fires a 360-degree scan, creating a dense point cloud: millions of individual distance measurements rendered as a three-dimensional map of every surface in range, every ridge and valley and sag captured in data points so numerous they'd take a human surveyor weeks to replicate by hand, assuming the human didn't quit first. Machine learning algorithms then segment that cloud, automatically separating concrete from walls, rebar chairs, and the finisher's lunch cooler. Automated analysis calculates flatness and levelness per ASTM E1155, the industry standard, and produces a report showing exactly where deviations live and how severe they are.

Under one minute from scan to report, with no chalk lines, no marking, no walking with a profiling tool, and full surface coverage rather than the sampled fraction a straightedge operator can manage in a day.

"With this digital tool, you can quickly get estimates for the foundation's smoothness with minimal human involvement," Hayes told ORNL's press office. "This enables faster building construction because FLAT can tell you within a minute if any mistakes were made and where they are so they can be corrected before the concrete hardens."

Before the concrete hardens. That's everything.

Wet Costs $100. Dry Costs $15,000.

Timing is what separates a clever lab demo from a tool that could actually save builders money, and in concrete work, timing operates on a merciless binary: wet or dry, fixable or expensive.

When concrete is wet, correcting a flatness deviation costs almost nothing: re-screed the low spot, bull float it, add material if necessary. Fifteen minutes, maybe thirty on a bad day, and call it $50 to $200 in labor for a task so routine nobody bothers writing it down.

Same deviation, after hardening, and the math gets ugly. Grinding runs $3 to $15 per square foot. Self-leveling compound costs $1.80 to $5 per square foot professionally applied. Slab jacking, polyurethane injection, or structural remediation for a foundation pushes into $4,500 to $37,500, per Angi's 2026 data. HomeLight puts full house leveling at $10,000 to $30,000.

Specific math on a realistic scenario. A 3/8-inch dip across 200 square feet of a 2,000-square-foot residential slab, caught wet, costs about $100 to re-screed. Caught dry: $600 to $3,000 in direct remediation, plus a week or more of schedule delay as the fix cures and trades play musical chairs around the damaged section, pushing your closing date further from the original timeline with every day that passes while nobody can touch that part of the floor.

FLAT doesn't prevent dips; finishers prevent dips. What FLAT does is guarantee every dip gets found in the 4-to-6-hour window when repair is trivially cheap instead of the 4-to-6-week window when it becomes a warranty claim, an insurance call, and a fight about who poured the northeast corner too thin at 4:45 on a Friday.

Current Methods Are Embarrassing

Here's how residential foundation testing actually works. A concrete technician lays a 10-foot aluminum bar across the slab surface and checks for gaps underneath, per ACI 117. Allowable gap is 3/4 inch from design elevation, which sounds precise until you learn the technician picks where to place the bar. ORNL's own research paper noted the straightedge method is "inexact and rarely representative of the entire floor since the technician is free to choose any location on the floor to perform the measurement."

Translation: the test checks wherever the operator feels like checking. On 2,000 square feet, a conscientious finisher might sample 30 to 40 locations over three hours, leaving 1,800-plus square feet evaluated by gut, by experience, and by the way the light catches the surface when you squint at it from across the pour, which is roughly as scientific as checking your engine oil by listening to the idle and nodding thoughtfully.

ASTM E1155 exists and produces actual F-numbers: FF for flatness, FL for levelness, on a scale where higher is flatter. Conventional slabs target FF20/FL15, carpet needs FF25/FL20, thin-set tile demands FF35/FL25, an established standard widely referenced in commercial work but almost never used in residential, because testing takes hours, requires specialized equipment, and costs $500 to $1,500 per test when contracted out. Most builders don't spec F-numbers at all, just the straightedge, and nobody complains until the floor installer shows up with a laser level and starts making phone calls.

FLAT brings F-number testing down to a laser scan. Whether that cost is justified is harder to answer.

Nobody Has Published This Math

ORNL hasn't released a cost-benefit analysis. Matters a lot, because the tool's value rests entirely on whether early-detection savings justify hardware cost, and right now we're estimating from adjacent data rather than measuring directly.

Mid-range terrestrial laser scanner runs roughly $15,000 to $40,000, with the FARO Focus Swift around $20,000 and Leica and Trimble units spanning that range depending on features. FLAT's software would presumably carry a separate license; ORNL lists the technology as available through its partnerships office.

Builder's calculation: fifteen foundations a year, one flatness issue caught annually that would have become a $3,000 remediation without early detection, and the scanner pays for itself in five to seven years. One issue that would have become a $15,000 structural claim instead, and payback is year one. Whether those catches actually happen depends on how often foundation flatness causes costly rework in residential, and nobody has reliably published that number. Foundation repair is a $4.9 billion annual U.S. industry, but that figure includes settling, cracking, water intrusion, and a dozen failure modes unrelated to original-pour quality, making it useless for isolating the specific savings FLAT targets without a dataset that apparently doesn't exist yet.

Honest answer: FLAT probably works economically for a custom builder doing $500K-plus homes with engineered flooring specs where a single rework event is catastrophic. Production builders running slabs at $180 per square foot with LVP flooring and no flatness spec probably aren't there yet, but give it three years and a scanner price drop.

Concrete Is Getting Smarter Everywhere

FLAT isn't alone in trying to drag concrete inspection out of the last century. A team published in Nature Communications demonstrated piezoelectric sensors embedded directly in concrete that monitor curing strength in real time, using deep learning to interpret electromechanical impedance signals from tiny transducers riding inside the slab as it hardens, reporting continuously on hydration progress and whether the mix is approaching design strength or falling behind, with prediction accuracy within approximately 15% of standard ASTM C39 cylinder break tests and the underlying sensing principle already incorporated into AASHTO T412, a new national standard for non-destructive testing that represents one of the first times AI-driven structural monitoring has crossed from research into codified engineering practice.

Highway-validated only, not residential, but picture what it replaces. A builder pours a slab, makes test cylinders, ships them to a lab, waits 7 days for an initial break and 28 days for final strength confirmation. Cylinders fail, and the slab's already holding framing. Embedded sensors would change the conversation entirely: hit 3,000 psi at hour 19, strip forms safely, cold snap slowed hydration by 11%, push the schedule two days rather than guessing.

Both tools share a thesis worth stating plainly. Concrete work is monitored with instruments designed before GPS, before cell phones, before most of the people swinging the straightedge were born. Better data exists. It's sitting in labs and journals. Residential construction hasn't noticed.

What You Can't Buy Today

FLAT is research, not a product. ORNL lists it for licensing ([email protected], 865-574-1051) and plans further demonstrations on suspended slabs, floor decking, and pier foundations. Nobody has commercialized it, and nobody has announced plans to.

What you can do right now: hire a concrete scanning service. Companies already use terrestrial laser scanners for floor flatness testing in commercial construction, typically $500 to $1,500 per visit, though without FLAT's machine-learning segmentation and automated reporting, analysis takes longer and costs more. Core capability exists, though you have to request it, and your concrete contractor will look at you like you asked him to wear a lab coat.

Same pattern, every time. A tool catches catastrophic problems early, proves itself in commercial or in a lab, and residential won't touch it until insurance forces the issue or a plaintiff's attorney waves the report at a jury and asks why nobody scanned the surface holding up a $900,000 house, at which point every builder in the state will discover they always believed in laser scanning and can't imagine how anyone ever worked without it. Straightedges aren't going anywhere. They don't need software licenses or firmware updates or annual calibration certificates from Oak Ridge. But next time a floor installer tells you your slab is out of spec and the fix will delay closing by three weeks, remember: 47 seconds and $100, or twelve weeks and $15,000. Pick.

Limitations

FLAT has been demonstrated at two housing developments, a small validation sample for efficiency claims of this magnitude. ORNL's reported 90%-plus time reduction is self-reported and has not been independently verified. Cost-gap analysis in this article uses foundation repair industry pricing; no published study isolates how often original-pour flatness defects, rather than settling or soil movement, cause residential claims. Piezoelectric concrete monitoring has been validated only on highway projects. Scanner prices are approximate; manufacturers negotiate volume discounts and don't publish consistent list pricing.