Drone Construction Survey and Progress Monitoring in India

Geocartis surveys active construction sites across India: pre-construction baseline, earthwork quantities, monthly or fortnightly progress capture, as-built comparison against design, and a dated 3D record of the site through time. Capture is at 5 cm ground sample distance or better, tied to permanent ground control that survives the whole project, with an RMSE report against every epoch. We are based in Ahmedabad and work nationally.

We are a survey and geospatial services company. We do not build drones and we do not sell software.

One flight is a photograph. A series is an asset.

A single drone survey of a construction site produces an impressive image and not much else. The value on construction work is almost entirely in repetition.

Fly the same site every month, registered to the same control, and three things become possible that a one-off survey cannot deliver.

Quantities between any two dates, computed from measured surfaces rather than estimated from cross sections at intervals.

Deviation from design, shown as a coloured surface across the whole site rather than as a list of discrepancies somebody found by inspection.

A dated record, which is what settles an argument two years later about when something was built, what the ground looked like before it, and whether an obstruction was actually there on the date claimed.

The third one is the sleeper. Most clients buy the first two and end up valuing the third most, because construction disputes are almost always about chronology, and chronology is exactly what a monthly georeferenced record establishes.

The baseline flight

The most valuable capture on any project is the one before the first machine moves, and it is the one most often skipped because at that point the site looks like nothing and nobody sees the need.

Original ground level along the whole site, measured and controlled, is the surface every earthwork quantity for the rest of the project will be computed against. Without it, quantities get argued from design sections, borrow records and whatever levels the contractor took with a level and a staff in the first week.

We push for this on every project, and we push for it before mobilisation rather than after clearing has started. Once topsoil has been stripped from part of the site, the baseline is already compromised for that part, and the argument about how much was stripped begins.

Alongside the surface, the baseline flight documents existing conditions: trees, structures, boundary features, access, overhead lines, water bodies, neighbouring construction and encroachment. On a site where the neighbour later claims damage, a dated pre-start record of their building is worth more than the whole rest of the survey programme.

Earthwork tracking

Earthwork is measured, certified and paid, and it is the single largest source of quantity dispute on Indian construction sites.

The traditional method is cross sections at fixed intervals with interpolation between them. Interpolation error on undulating ground runs in both directions and is invisible, because there is no data between the sections to contradict it. On a site with irregular ground, the error is not small.

A surveyed surface removes the interpolation. Cut and fill are computed from measured ground, and because the underlying surface exists, quantities can be recomputed at any section spacing afterwards. If a section interval is disputed, you re-run it rather than re-survey.

What we deliver on earthwork:

Cut and fill volumes between any two epochs, or against the design surface, reported by zone, grid or chainage segment so they reconcile against your bill of quantities.

Stockpile and borrow area volumes, so material brought to site can be reconciled against material placed.

Layer-by-layer placement measurement on embankments, which supports both payment and compaction records.

Mass haul information, showing where material moved from and to across the site.

The base surface is declared in every report. Volume is the difference between two surfaces, and the base is a choice: original ground, design formation, a previous epoch, or a defined datum plane. Two competent surveyors produce different volumes from identical data by choosing differently. A volume without a stated base is not a measurement.

As-built comparisons

This is where drone survey gets oversold most often, so it is worth being precise about what it does.

We register the survey to your design and produce a deviation surface: where the built condition sits above or below the design, mapped as a continuous coloured surface with values, rather than as a set of spot checks.

For earthworks, formation levels, platform levels, road profiles, embankments, slopes, yard surfaces and large flat elements, that comparison is genuinely useful and the accuracy is appropriate to the tolerance. A formation level specified to 25 millimetres and a survey resolving to 30 to 50 millimetres vertically are in a sensible relationship. You will catch a systematic problem across an area, which is the failure mode that costs money.

The coordinate problem, which nobody warns you about. Your design sits in a project grid with an arbitrary origin, a project north and often a project datum for levels. Our survey sits in UTM WGS84 with orthometric or ellipsoidal heights. Comparing them requires an exact transformation between the two, and getting it approximately right produces a deviation surface that looks plausible and is wrong everywhere by a constant amount.

We establish that transformation explicitly, from surveyed points that exist in both systems: your setting-out stations, your temporary benchmarks, your grid intersections. If those do not exist or cannot be located, we say so before quoting the comparison, because an as-built comparison without a verified transformation is worse than no comparison at all. It is a confident wrong answer.

The tolerance limit, stated plainly. Photogrammetry at 5 cm GSD resolves vertically to roughly 10 to 15 centimetres and horizontally to 5 to 10. Structural tolerances for concrete and steel are in millimetres. A drone survey cannot verify column verticality to code tolerance, cannot check beam camber, cannot confirm cover to reinforcement and cannot certify a floor slab to flatness specification. Those are total station, level, plumb and specialist instrument work.

What the survey does is find the problems that are big enough to see and spread over enough area to matter. A platform built 200 millimetres low across half a hectare. A slope built at the wrong angle over 100 metres. A road formation with a systematic cross-fall error. Those are the ones that cost real money, and they are exactly the ones a spot-check regime misses.

Tasks and schedules

Clients increasingly want the survey linked to the programme, and there is a useful version of this and an oversold version.

What a survey epoch can establish about an activity: whether it has visibly started, whether it appears visually complete, and how much of a measurable quantity has been done where the activity has a measurable quantity. Excavation, filling, concrete pours by area, floors constructed, structures erected, road length laid, area paved.

What it cannot establish: whether work is complete to specification, whether hidden work was done correctly, whether reinforcement was placed as drawn, whether curing was adequate, or the true percentage complete of an activity whose progress is not physically visible from above.

Percentage complete, in general, is inference rather than measurement. A structure that looks eighty percent built from the air may be forty percent complete in value terms, because services, finishes and fit-out follow. We are explicit about which figures in our reporting are measured and which are assessed, and we do not present a visual assessment as a measurement.

Used honestly, the linkage works well. We tag survey observations to your programme activity codes, so each epoch produces a status against the activities that are physically visible, alongside the measured quantities. The output goes to your planning team as a structured dataset rather than as a report they have to re-key, and it gives the planner an independent check against the site’s own progress claim.

On a delayed project, the value flips. A monthly series showing exactly when each work front started and stopped is the evidentiary spine of any extension of time claim or defence.

The sectors we work in

Building construction. Residential towers, commercial buildings, mixed-use developments and townships. Baseline and excavation quantities, basement and raft progress, structure rising, facade and roof condition, site logistics and material stacking, and a dated record for the promoter’s own file. Where the project is registered under RERA and the promoter has quarterly reporting obligations to the state authority’s portal, a quarterly capture produces the visual and measured basis for that update rather than a photograph taken from the site office roof.

Warehouse and industrial construction. This is where drone survey performs best of all, and the reason is geometry. A warehouse is a very large, very flat, single-level structure, which is exactly the shape photogrammetry measures well. Pad levels across hundreds of thousands of square feet, floor slab area poured, roof sheeting progress, dock and apron levels, yard paving, and the surrounding earthworks. On a large logistics park the survey covers ground a walking crew cannot cover weekly.

Road construction. Corridor progress, earthwork by chainage, layer placement, structure progress at bridges and interchanges, borrow area depletion and diversion layouts. National Highway projects also carry NHAI’s monthly drone video requirement, which we cover in more detail on our road survey page and which is usually worth flying alongside the measurement capture rather than separately.

Bridge construction. Foundation and pier progress, deck erection, approach embankments and the surrounding site. Bridges need oblique capture as standard, because a nadir survey of a bridge gives you the deck and almost nothing of the piers, bearings, abutments or soffit. For a bridge over water, the exposed structure is measurable from the air and anything below water level is not.

Metro construction. Viaduct and pier progress along the alignment, casting yard output and segment stockpile counts, station box excavation and structure, depot construction, and traffic diversion layouts. Metro work sits in dense urban corridors, which makes the airspace and safety planning harder than the surveying, and casting yards are one of the neatest applications we have because segment counting and stockpile measurement from the air is fast and unambiguous.

Railway construction. Formation earthwork by chainage, structure progress at bridges and ROBs, yard and station construction, and material stockpiles along the corridor. Work on or adjacent to operational railway land needs permission from the railway administration in addition to DGCA clearance, and the overhead equipment and live-track constraints are covered on our railway survey page.

Ship construction and shipyards. Yard layout, block storage and movement, slipway and dry dock area, fabrication shop surroundings, stockyard measurement and construction of yard facilities themselves. A frank caveat belongs here: most shipyards in India of any scale are either defence establishments, adjacent to one, or inside a port perimeter, and a substantial number sit in restricted airspace where no commercial operator can fly regardless of what the yard permits. We check the airspace position before quoting rather than after, and on many yards the answer will be that the work cannot be done from the air.

How we do it

Step 1: Setting up the programme, not the flight

Construction survey is bought as a series, so the first conversation is about the series. Frequency, duration, what gets measured each time, what gets reported, and to whom.

Monthly suits most building and infrastructure projects. Fortnightly suits earthwork-heavy phases where quantities move fast. Quarterly suits a promoter’s reporting cycle. Some projects want weekly during a critical phase and monthly otherwise, and that is a sensible way to buy it.

We also agree the reporting format at the start, matched to how your team already works: your zone or block naming, your chainage, your activity codes, your bill of quantities structure. A survey delivered in our structure creates a re-keying job for somebody.

Step 2: Permanent control that survives the project

This is the decision with the longest consequences on a construction project, and it is made in the first week.

A construction site is the most hostile environment for survey control there is. Everything moves. Points get built over, buried under stockpiles, knocked out by plant, or removed by somebody who did not know what they were.

We establish permanent control outside the works, on ground that will not be disturbed for the life of the project: boundary walls, adjacent stable ground, existing structures, purpose-set pillars in agreed protected positions. We survey them with DGPS, document them with ground-level photographs and coordinates, and re-occupy them at every epoch.

Where the project has an existing setting-out grid and temporary benchmarks, we tie into them and establish the transformation between the project system and the survey system at the outset, before it is needed for an as-built comparison. Working that out later, from a site whose original stations have since been built over, is a much harder job.

Independent checkpoints go in and are withheld from the adjustment, and we report RMSE at every epoch rather than once at the start. On a series, drift between epochs matters more than absolute accuracy in any one of them.

Step 3: Airspace, and the site’s own rules

We check the site against the Digital Sky airspace map before quoting. Green zone to 120 metres above ground level without prior flight permission, yellow zone with permission and a lower ceiling, red zone not at all. Urban construction sites, metro corridors and anything near an airport are frequently yellow.

Our aircraft carry active Unique Identification Numbers and our pilots hold DGCA Remote Pilot Certificates.

Then the site’s own requirements, which on a well-run project are substantial: safety induction, permit to work, PPE, notification to the crane operators, and agreement on when flying happens. We plan for those rather than treating them as an obstacle, and on a recurring programme they only need setting up once.

Step 4: Flight planning on a live site

Ground sample distance is pixel pitch multiplied by flying height divided by focal length. A 20 megapixel one-inch sensor with 5472 pixels across 13.2 millimetres has a pixel pitch of about 2.41 micrometres. With an 8.8 millimetre lens at the 120 metre ceiling that gives roughly 3.3 centimetres GSD.

Flight plans on a monitoring programme are repeated, not redesigned. Same lines, same altitude, same camera settings, same time of day where possible, every epoch. Consistency between epochs is worth more than optimising any single flight, because the deliverable is the difference.

Obliques are standard. A nadir block gives you ground and roofs. Every vertical element on a construction site, structure faces, retaining walls, pier shafts, deck edges, formwork, is reconstructed properly only from obliques. On a rising building the nadir view becomes progressively less useful as the structure gets taller.

Rising structures change the flight plan. A site flown at 120 metres at baseline may have a 60 metre structure on it by month eighteen, which halves the clearance and doubles the GSD variation across the site. We revise the plan as the structure rises, and note the revision in the epoch report so the change is visible rather than silent.

Step 5: Capture

The tower crane is the hazard that matters. A crane slews through a large radius, the jib is at working height, and the hook block and lifting gear are hard to see. Flight plans mark crane positions and slew radii as exclusion volumes, and the crane operators are told when we are flying. On a site with multiple cranes this is coordinated with the site engineer rather than judged from the ground.

Other constraints: people below, which means take-off and landing away from work areas and no transit over occupied zones; overhead cables around and across urban sites; dust from earthmoving, which degrades imagery; and the sheer amount of movement, since plant and vehicles in the imagery produce ghosting and spurious points.

Timing is scheduled for high sun, and on a monitoring series for the same time of day each epoch, because shadow position changes what is visible in narrow spaces between structures.

Step 6: Processing and comparison

Aerial triangulation with bundle adjustment against the control, camera self-calibration, dense matching, classification.

Classification on a construction site is unusually hard, and it is worth knowing why. The classifier has to separate finished ground from stockpiles, formwork, scaffolding, stacked material, plant, site huts and temporary works, and several of those have the geometry of the thing being measured. A stockpile looks like a landform. A stack of precast segments looks like a structure. These get separated by hand against the site layout and against what the site team tells us is where.

Comparison against the previous epoch and against design then runs on the cleaned surfaces, clipped to the common measured area so that differences reflect real change rather than differences in coverage.

Step 7: Reporting

Each epoch produces the same report structure, which is what makes a series readable: orthomosaic, surface models, quantities by zone against the previous epoch and against design, deviation surface where an as-built comparison is in scope, activity status against your programme codes, the measured RMSE for that epoch, and the flight record.

Consistency across epochs is the point. A series where the reporting format changes month to month is a series nobody can read.

Accuracy on a construction site

Ground sample distance and accuracy are different quantities. GSD is the ground area one pixel covers. Accuracy is how close a coordinate is to the truth.

For a well-controlled block, horizontal RMSE typically lands between one and two times the GSD and vertical between two and three times. At 5 cm GSD that suggests roughly 5 to 10 centimetres horizontal and 10 to 15 centimetres vertical. Every epoch report carries the measured figures.

Two things are specific to monitoring work.

Relative accuracy between epochs is what matters, and it is better than the absolute figure provided both epochs sit on the same control. This is the argument for permanent control. Two surveys at 12 centimetre RMSE on shared control produce a far better difference than two surveys at 8 centimetres on independently established control.

Systematic error dominates volume. Random noise across a large surface largely cancels when integrated into a volume. A systematic vertical bias does not, it multiplies by area. A 15 centimetre bias across a five hectare site is 7,500 cubic metres of apparent material, which on a paid earthwork item is real money.

Occlusion is the main degrading factor. Ground beneath scaffolding, under a slab, inside a structure or behind a stockpile is not visible from above, and interpolating across it produces a surface that looks complete and is not measured. We mark occluded areas rather than filling them silently.

Deliverables and output file formats

Raw imagery is handed over with each epoch. On a project where the record may be produced in a dispute years later, holding the raw data yourself matters.

Photogrammetric marking

A related service worth naming separately, because clients ask for it and do not always know what to call it.

Photogrammetric marking is precise identification and annotation of built features against the survey: gridline positions as built, column and pier locations, floor edges, opening positions, wall lines, plinth and slab edges. The output goes to the site team as vector layers over the orthomosaic and the point cloud, in DXF, so it lands in the drawing environment they already work in rather than as a report.

On a repeat programme it produces a running record of where things actually ended up, which is the input to any as-built drawing set and to any argument about setting-out.

Where we work

We are based in Ahmedabad and work nationally. Construction monitoring is a recurring commitment rather than a single visit, so we structure it as a programme with scheduled mobilisation.

Gujarat. Ahmedabad and Gandhinagar including GIFT City, the Ahmedabad Metro corridors, the Sabarmati and Ahmedabad high speed rail station areas, Dholera SIR, the industrial and warehousing belt through Sanand, Changodar, Bavla and Kheda, and the Ahmedabad-Vadodara and Ahmedabad-Rajkot corridors. Plus Surat, Vadodara, Rajkot, Bharuch, Ankleshwar, Jamnagar, Mundra and Hazira.

Maharashtra and the west. Mumbai, Navi Mumbai, Thane, Pune, Pimpri-Chinchwad, Nashik, Nagpur and Chhatrapati Sambhajinagar.

Northern India. Delhi NCR including Gurugram, Noida, Greater Noida, Faridabad and Ghaziabad, plus Jaipur, Chandigarh, Ludhiana, Dehradun, Lucknow, Kanpur and Patna.

Central India. Indore, Bhopal, Jabalpur, Raipur and Nagpur.

Eastern India. Kolkata, Bhubaneswar, Ranchi, Jamshedpur and Guwahati.

Southern India. Hyderabad, Bengaluru, Chennai, Coimbatore, Vijayawada, Visakhapatnam and Kochi.

What we do not do

We do not verify structural work to code tolerance. Column verticality, beam camber, cover to reinforcement, slab flatness and similar checks are total station, level and specialist instrument work in millimetres. Photogrammetry works in centimetres.

We do not certify quality, workmanship or compliance. We measure geometry and record condition.

We do not inspect hidden or covered work. Once concrete is poured, what is inside it is beyond any camera.

We do not survey building interiors. Terrestrial laser scanning is the method for internal as-builts.

We do not measure below water on marine and bridge work.

We do not carry out setting out. We record where things are, not where they should go.

We do not fly in red zones, which on shipyard and port work rules out a meaningful proportion of sites.

We do not fly over occupied work areas or within crane operating volumes.

Frequently asked questions

How often should we fly? Monthly suits most building and infrastructure projects. Fortnightly during heavy earthwork phases where quantities move fast. Quarterly where the driver is a reporting obligation. The right answer depends on how fast your quantities change and what the survey is feeding.

Can you measure our earthwork quantities for billing? Yes, from measured surfaces against a declared base, reported by zone or chainage so they reconcile with your bill of quantities. Whether your client accepts drone-derived quantities for certification is a contractual question worth settling before the survey rather than after, and it is much easier to settle when a baseline survey exists.

Can you compare what we built against the design model? Yes, provided we can establish an exact transformation between your project grid and the survey coordinate system, from points that exist in both. Without that, the comparison is a confident wrong answer and we will not produce it. For earthworks, formation and large flat elements the comparison is genuinely useful. For structural tolerance it is not the right instrument.

Can you tell us our percentage complete? We can give you measured quantities, and visual status for activities that are physically visible from above. Percentage complete in value terms includes services, finishes and fit-out that a camera cannot see. We label measured figures and assessed figures differently, and we do not present one as the other.

Can you fly over an active site? Yes, subject to DGCA airspace clearance and your site’s safety requirements. We plan around crane slew radii and occupied work areas, take off and land away from work fronts, and notify crane operators before flying.

What about our shipyard? Check the airspace first. Most Indian shipyards of scale are defence establishments, adjacent to one, or inside a port perimeter, and many sit in airspace where no commercial operator can fly. We will tell you the position before quoting.

Do you handle NHAI monthly drone videography? Yes, and on a highway package it makes sense to fly the video and the measurement capture together rather than mobilising twice. Details are on our road survey page.

We are already six months into the project. Is it too late? No, but you have lost the baseline. We can start a series from today and everything from here is measurable. Quantities before today will still be argued from whatever records exist. Start now rather than at month twelve.

What file formats do you deliver? Orthomosaic as GeoTIFF and JPEG. Surface models as GeoTIFF. Point cloud as LAS, LAZ and E57. Mesh as OBJ or FBX. Drawings and marking as DXF and DWG. Sections as DXF, DWG, PDF and XLSX. Quantity and progress reports as PDF and XLSX.

Can the data go into our BIM environment? Point clouds in E57 and LAS import into the common platforms, and mesh models in OBJ. The registration to your project coordinate system is the part that needs setting up properly, and it is the first thing we do rather than the last.

Start a monitoring programme

Send us the site boundary as a KML or shapefile, the project duration, the frequency you want, and what the survey needs to feed: quantities, progress reporting, as-built comparison, or a record for the file. Tell us whether a setting-out grid and benchmarks exist, and whether the site has started.

We will return the airspace position, the control plan, the epoch schedule, the reporting format and a price for the programme rather than per visit.

If the site has not broken ground yet, the baseline flight is the one to book first.