Industries / Construction and property / Real estate / Home inspection
Moisture and pests
This work looks for water where it should not be, for the damp, mold and rot it causes, and for wood-destroying insects, and traces each to a cause where the cause can be seen.
4
Milestones achieved, from 2005 to 2019.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2005
Sealed crawl spaces stay drier
A crawl space is the low space under a house built off the ground. Codes called for vents in its walls on the idea that outside air would carry moisture away. From 2001 Advanced Energy, a North Carolina energy group, compared 12 matched houses on one cul-de-sac in Princeville, North Carolina, with US Department of Energy (DOE) funding, and reported in June 2005. In summer the closed crawl spaces, with their vents sealed and a sealed plastic liner, stayed generally below 60% relative humidity, while the vented ones were often above 80%; the closed ones also cut heating and cooling energy. DOE's Building America program says this research influenced the 2009 and 2012 International Residential Code (section R408.3), which allows unvented, conditioned crawl spaces.
LessonIn a humid climate, outside air brings more moisture into a crawl space than it takes out.
Credit
Bruce Davis and colleagues at Advanced Energy; US DOE
2007
Meters cannot prove rising damp
Rising damp is ground water drawn up through a masonry wall. On 12 October 2007 BRE, Britain's Building Research Establishment, published a revised Digest 245 on diagnosing it, written by Peter Trotman. It says salts build up in the walls of old buildings even without a damp problem, so high readings from a moisture meter alone are not conclusive, and it describes drilling samples from the wall to measure moisture and salt content. In 2020 the Property Care Association, the UK trade body for damp specialists, wrote that electrical meters "can give rise to confusion and incorrect diagnosis" where salts are present, and that oven-drying drilled samples is "by far the most accurate method".
LessonA surface meter shows where to look, not how much water there is or where it came from.
Credit
BRE; Peter Trotman
2009
WHO sets no mold limit
The World Health Organization's guidelines on dampness and mould, published by its Europe office in 2009, found sufficient evidence that people in damp or moldy buildings face higher risks of respiratory symptoms, respiratory infections and worse asthma. They also found that the links could not be measured precisely enough to set any safe level of mold, so they recommend preventing and fixing dampness instead. The US Environmental Protection Agency (EPA) says there are no federal limits for mold or mold spores, and that where mold can be seen, sampling is usually unnecessary.
LessonA house report on mold is a report on moisture and visible growth, not on a spore count.
Credit
WHO Regional Office for Europe; US EPA
2019
Termite form drops old clues
Form NPMA-33 is the National Pest Management Association's (NPMA) report of a wood-destroying insect (WDI) inspection, the termite check made for many US home sales, and it can go to the buyer's lender; California's pest board makes its own form, 43M-41, the official report and allows NPMA-33 only as an attachment. In July 2019 NPMA released a revised NPMA-33 after nearly two years of review by its Wood Destroying Organisms committee. The new form no longer requires inspectors to report "visible evidence of possible previous treatment", because bait stations and spot treatments can leave no trace. NPMA said what matters more is whether the house is covered by an ongoing termite program run by a licensed firm.
LessonA form should ask for what an inspector can still see, and get the rest from records.
Credit
NPMA Wood Destroying Organisms committee
Watch for
Photos screened for water damage
Inspection Support, which makes home inspection software, says its Defect Detector scans an inspector's photos, flags defects such as cracks, water damage and mold, and writes comments to paste into the report, for up to 100 images a day per inspector. Its help page gives no accuracy figures.
Achieved whenan outside test on photos whose moisture was confirmed by meter or by opening the wall reports what share of real damage the tool finds and what share of its flags are wrong.
Claimed by
Inspection Support, September 2026
Thermal scans for termites
Guide Sensmart, a maker of thermal cameras, wrote in July 2026 that thermal cameras cannot see through walls but can show temperature patterns "that may be related to termite activity", to be confirmed with other tools. A University of California, Berkeley study for California's Structural Pest Control Board found that infrared has imaged damage by subterranean termites, which build with wet mud, but that drywood termites need no water and so leave no temperature difference to image.
Achieved whena field study that opens the flagged walls reports how often a thermal pattern marked live termites, by termite type.
Claimed by
Guide Sensmart, July 2026
Open
01
Wet or dry, by meter
A public set of indoor thermal images of house walls and ceilings in which every cool or warm spot was checked on the spot with a pin moisture meter and labeled wet or dry, with the cause noted where it was found; then a measure of how often inspectors and AI models read the images correctly.
The briefCredit · unclaimed02
Robot crawl against human crawl
A published tally that compares what a remote-driven crawl space robot finds with what an inspector finds crawling the same space: standing water, wet or rotted wood, fallen insulation, pests and structural faults.
The briefCredit · unclaimed
Open 01 · the brief
Wet or dry, by meter
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public set of indoor thermal images of house walls and ceilings in which every cool or warm spot was checked on the spot with a pin moisture meter and labeled wet or dry, with the cause noted where it was found; then a measure of how often inspectors and AI models read the images correctly.
Why it is still open
A builder quoted when the FLIR ONE came out warned that thermal images are easy to misread. The public labeled sets of house thermal images found in this research are about heat loss, not water: a 2023 Scottish set of 1,800 annotated images of heat-loss areas, and a 2024 Toronto study of drone images of building envelopes. This research found no public set labeled with moisture readings.
Built with what exists
A phone thermal camera such as a FLIR ONE, or a handheld FLIR or Testo camera; a pin moisture meter; images taken during inspections with the owners' consent; Label Studio for labeling; open vision models fine-tuned in PyTorch, and at least one commercial AI model for comparison.
First step
Photograph and meter 50 thermal anomalies in 10 houses during routine inspections, with the owners' consent.
Done looks like
At least 1,000 images from at least 100 houses, each spot with its meter reading and a wet or dry label, released under an open license, and the share of wet spots and of dry look-alikes that five inspectors and two AI models call correctly.
Open 02 · the brief
Robot crawl against human crawl
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A published tally that compares what a remote-driven crawl space robot finds with what an inspector finds crawling the same space: standing water, wet or rotted wood, fallen insulation, pests and structural faults.
Why it is still open
Companies such as SuperDroid Robots sell crawl space robots to home inspectors. InspectorPro, an insurer of home inspectors, wrote that the robots may lose signal or have limited views, that "the crawlspace crawler could miss important defects", and that there is little claims data on them. This research found no published comparison of robot and human findings.
Built with what exists
A commercial crawl space robot with lights and a camera; a phone thermal camera and a pin moisture meter for the person; one written checklist of defects used by both; a spreadsheet for the tally.
First step
Agree with two inspection firms to run the robot first, then crawl, on their next ten crawl space inspections.
Done looks like
30 crawl spaces, each inspected first by robot and then by a person who has not seen the robot's footage, with every finding listed and a published tally of what each method found and missed.