Field notes · Measurement, method and limits

Articles

Working notes from a drone operator who measures his own error. Every number here came off a real flight, and the results that went badly are published with the ones that went well — because the second kind is only worth anything next to the first.

9 Aug 2026
Measurement
~10 min read

How accurate is a sub-250 g drone, really?

Almost every operator in this market claims "accurate to 2 cm" and almost none of them have flown the same site twice to check. I did: the same car park, the same saved mission, 24 hours apart, with nothing changed in between — so everything the model reported was error. The result, the method, the surfaces it does not apply to, and the three findings that don't flatter the kit, including a volume over-read of 2.7× and an automated change detector that missed a box entirely.

Read the study →

90% within: 0.237 m
95% within: 0.311 m
Hardstanding scatter: 0.033 m
Volume over-read: 2.7×

What's coming

Written when the work behind them is done and measured, not before:

Correcting the per-flight datum shift — whether 14–24 cm of constant offset can be removed, and what is actually left when it is
Volume error at stockpile scale, on something considerably larger than a shoebox
What a month of progress packs looks like on a live site, with the client's permission
Flying inside the Heathrow zone: where the restrictions bite around Slough, Windsor and the M4, and where they don't

The day job

Monthly construction progress monitoring, from Slough.

The same mission flown every visit — orthomosaic, labelled stills and a month-to-month comparison, on your desk by the third working day, before your progress meeting. Berkshire, Bucks, the M4 corridor and Greater London.

Get a quote How it works Capability statement