How to Check As-Built Work Against the Design on a Drone Map

Written by
Kayley Greenland
Last updated:
November 12, 2026

To check as-built work against the design, lay the current drawing revision over a recent, well-controlled drone map, fade it, and walk the site element by element looking for anything missing, moved or extra. An overlay checks layout in plan. Levels, tolerances and anything buried still need a survey.

‍

Key takeaways

  • Bad information is expensive. According to FMI and PlanGrid, miscommunication and poor project data account for 48% of rework on US construction jobsites.
  • The wrong revision wastes the check. The same study found team members spend five and a half hours a week hunting project data, such as revised drawings.
  • An overlay answers one question well: is it where the drawing says? It doesn't check levels or thickness.
  • The map sets the ceiling. One 2026 study found horizontal error of about 0.06 m with ground control and 3.5 m without.
  • Measured checks are possible. A 2023 KU Leuven study compared drone-derived road elements with the design model element by element, against a manual check.

‍

What can a drone map overlay tell you about as-built work?

A design overlay tells you whether what was built sits where the drawing says, in plan. It shows elements that are missing, out of position, the wrong shape or not on the drawing at all. It can't tell you levels, thickness or anything buried, and its accuracy is the map's plus the alignment's.

That is still a lot. A curb line that drifts off its alignment, a pad built in the wrong spot, a pit or a footing that never went in, a haul road cut across a setback: all of these show up the moment the drawing sits over the capture. So do things the drawing never asked for, such as a stockpile parked on a future building footprint.

The same check works beyond construction sites. On a quarry or a mine, the drawing might be a pad layout, a haul road design or a bund line, and the question is the same: does the ground match the plan?

‍

What do you need before you compare?

You need the current drawing revision, a drone map captured after the work you want to check, and a map you can trust: ideally one captured with ground control. The drawing must sit on the map, either from its own coordinates or by lining it up with matched points.

Start with the revision, because it is the easiest thing to get wrong. According to the 2018 Construction Disconnected study by FMI and PlanGrid, construction team members spend five and a half hours a week hunting down project data, such as revised drawings. A check against a superseded drawing is precise and useless.

Then the capture. It should be recent enough to show the work in question, and captured the same way as the last one, so two checks can be compared. If the drawing has no coordinates, see our guide to line up a site plan without coordinates first.

‍

How do you check as-built work against the design, step by step?

Put the current revision over the latest capture, check the alignment at known features, fade the drawing, and walk the site one element at a time. Mark every discrepancy where you find it, with the date and the revision. Send anything that matters to the surveyor to confirm before anyone acts.

  1. Confirm the revision. Check the drawing number and revision against the current register.
  2. Use a fresh capture. Make sure the map postdates the work you're checking.
  3. Check the alignment. Before you judge the site, check the drawing against features you know are right, such as an existing building corner.
  4. Fade the drawing. Set it to around 50 to 70% opacity, so the design and the ground both show.
  5. Walk it element by element. Curbs, pads, pits, footings, roads, then anything the drawing doesn't show.
  6. Mark what's off. Drop an annotation on each discrepancy and measure the offset where it matters.
  7. Record what you compared. Note the capture date and the drawing revision with the result.
  8. Hand it to the surveyor. Anything that could cost money or a rebuild gets confirmed on the ground.

Step 3 matters more than it looks. If the drawing is out at one corner, everything near that corner will look wrong, and the work will be blamed for the alignment's error.

‍

How accurate does the drone map need to be?

Accurate enough that the error you're looking for is bigger than the map's own error. With good ground control, a drone map is reliable to a few centimeters. Without it, it can be meters out. An overlay is a strong screen for anything off by decimeters, and no substitute for a setout survey.

According to a 2026 study of a drone survey in Nepal, an orthophoto made with ground control points had a horizontal error (RMSE) of about 0.063 m. Without them, the error was about 3.507 m. The help center for Birdi (birdi.io) gives a similar picture for measurements on the map: around ±40 mm with RTK and ground control, and meter-level with GPS only.

The alignment adds its own error on top. A drawing lined up by hand is only as good as its control points, so check it at a feature you didn't use to place it.

So use the overlay to find problems, and the surveyor to settle them. Where a contract or specification sets a tolerance in millimeters, the overlay tells you where to look, and the survey tells you whether it passes. For handing drone data to survey and design software, see our comparison of as-built survey software.

‍

What can't an overlay show you?

An overlay can't show height. Levels, grades, pavement thickness and cut and fill against the design need surface data or a survey, not a drawing on a map. It also can't see inside a building, under the ground or under a roof, and it only shows what the drawing and the capture both cover.

Research makes the split clear. In a 2023 study in the ISPRS Archives, De Winter and colleagues at KU Leuven compared a road site in Belgium with its design model using an RTK drone. They handled the check in two parts: planimetry, the position of each element in plan, from drone images, and altimetry, the height, from a 3D model of the site. Their method matched a manual check closely in both.

Their starting point is worth noting too. They describe current checks as manual surveys with GNSS and total stations, which are "selective whereby a limited number of points are measured". A drone map covers the whole site, which is exactly why it is good at finding the element nobody measured.

‍

How often should you compare the site with the design?

Compare at every capture during the phases where layout is being set, and always before work is covered up: before a pour, before backfill, before a seal. A discrepancy found on the map that week costs a conversation. The same one found after the next trade has built on it costs a rebuild.

A steady capture rhythm makes each check faster, because the drawing is already on the map and the last result is beside it. Our guide to repeat drone surveys covers keeping captures consistent.

‍

How do you check as-built work in Birdi?

In Birdi, add the drawing to the map as an overlay: a georeferenced PDF is placed automatically, and a PDF, JPG or TIFF without coordinates is lined up by matching points. Then set the drawing's opacity, compare it with the latest capture, and mark discrepancies with annotations and measurements.

Multi-page drawing sets work page by page, and each page you pick becomes its own layer. A CAD export with vector data comes in as linework you can style, with individual CAD layers hidden. People outside the paid team can view the shared map as guests. The overlays themselves are new, on Plus, Growth and Ultimate plans: see Birdi's map overlays release for what changed.

‍

When is an overlay enough, and when do you need a survey?

An overlay is enough to find out whether work is in the right place, to brief a subcontractor and to decide where the surveyor should look. You need a survey when the answer has a tolerance, a level or a payment attached. Most teams need both, and the overlay makes the survey shorter.

If the people checking the site against the design aren't GIS specialists, Birdi is a sensible option for construction and mining teams that capture regularly, because the drawing, the capture and the markups share one map. However, a team that needs cut and fill against a design surface, or a 3D model checked against BIM, will need a tool built for that.

‍

Frequently asked questions

‍

What is an as-built check?

An as-built check compares what was built with what the design says should be there. It looks for elements that are missing, out of position, the wrong shape or not on the drawing. A drone map overlay does this in plan for the whole site; a survey confirms positions and levels where a tolerance applies.

‍

Can a drone survey replace an as-built survey?

Not where a tolerance or a legal record is involved. A drone map with good ground control is reliable to a few centimeters in plan and covers the whole site, which makes it a strong first check. A surveyor still confirms setout, levels and anything that needs a signed record.

‍

How accurate is a drone map for checking as-built work?

It depends mostly on ground control. One 2026 study found horizontal error of about 0.06 m with ground control points and about 3.5 m without. A drawing lined up by hand adds its own error, so check the alignment at a known feature before judging the work.

‍

Can you check levels from a design overlay?

No. A drawing on a map shows position in plan, not height. Levels, grades and cut and fill against design need surface data, such as an elevation model compared with a design surface, or a survey. Use the overlay to find where to look, then check levels there.

‍

Which drawing revision should you compare against?

Always the current one, confirmed against the drawing register before you start. A check against a superseded revision gives a precise answer to the wrong question. Record the revision and the capture date with every result, so anyone reading it later knows what was compared.

‍

About the author

This guide was written by the Birdi team. Birdi (birdi.io) is a geospatial AI workspace that construction and mining teams use to share drone, 360 and design data in one map. We read every source on November 11, 2026, and list each one below.

‍

Related reading

‍

Sources

‍

Research and standards

  1. FMI and PlanGrid. "New Research from PlanGrid and FMI Identifies Factors Costing the Construction Industry More Than $177 Billion Annually." PR Newswire, August 1, 2018. https://www.prnewswire.com/news-releases/new-research-from-plangrid-and-fmi-identifies-factors-costing-the-construction-industry-more-than-177-billion-annually-300689826.html
  2. De Winter, H., Bassier, M., De Geyter, S. and Vergauwen, M. "Phased Accuracy Analysis in Road Construction: Using BIM and Photogrammetric Output." International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLVIII-1/W2-2023, 1627 to 1633, 2023. https://doi.org/10.5194/isprs-archives-XLVIII-1-W2-2023-1627-2023
  3. Bisht, B. S., Subedi, N., Sapkota, R. K. and Thapa, L. D. "Comparative Accuracy Assessment of UAV-Derived Orthophoto, DSM, and DTM Generated with and without Ground Control Points: A Case Study of the LMTC Premises, Nepal." Journal of Land Management and Geomatics Education 8(01), 2026. https://doi.org/10.3126/jlmge.v8i01.97951

‍

Birdi

  1. Birdi Help Center. "How to add a georeferenced PDF or images to your map as a layer." 2026. https://help.birdi.io/en/articles/17751040-how-to-add-a-georeferenced-pdf-to-your-map-as-a-layer
  2. Birdi Help Center. "How to add a PDF or image without coordinates to your map." 2026. https://help.birdi.io/en/articles/18239994-how-to-add-a-pdf-or-image-without-coordinates-to-your-map
  3. Birdi Help Center. "What level of accuracy are the annotation tools against the map/model?" https://help.birdi.io/en/articles/9488667
  4. Birdi Help Center. "How to update a user role and permissions." 2025. https://help.birdi.io/en/articles/11091551
Kayley Greenland
Kayley is Birdi's product manager and is across all the nuts and bolts of our feature releases, product roadmap and platform vision.