What we have checked on real footage
A 4.01 GB GoPro MAX recording yielded 9,316 GPS measurements. JavaScript coordinates matched all coordinates in an independently generated Python reference. The metadata-only read touched 4,815,628 bytes of the 4,005,176,737-byte source. These are recorded engineering results from one file; storage speed, browser, camera mode and file structure can change performance.
A negative result matters too
A supplied 3.3 GB Insta360 X3 recording contained no supported GPS block. The parser reported that limitation instead of inventing a route. That verifies this negative case; it does not establish successful GPS extraction from every X3 recording or exclude undocumented metadata.
Synthetic tests and their boundary
Authored fixtures exercise GPS9 decoding, the documented Insta360 GPS layout, corrupt or truncated input, quality rejection and segment preservation. They help detect implementation regressions. They cannot replace recordings from real camera firmware or establish field accuracy.
How we check evidence and exports
Review testing checks source file identity, video time, GPS availability and the retained evidence frame. Coordinates inside a GPS gap remain unavailable. KML and GPX output is checked for parseable XML, while spatial exports preserve segment boundaries. Camera location and an independently placed object location remain separate.
A pilot acceptance checklist
- Agree the corridor, camera modes, file types and required deliverables.
- Keep the original recording and inspect whether it contains a GPS track.
- Compare selected positions against independently surveyed control if positional accuracy is a requirement.
- Check video alignment at several identifiable events, including the start and end.
- Review a known sample of findings and explicitly record missed or ambiguous evidence.
- Open exports in the receiving GIS or reporting workflow and reconcile counts, IDs and missing values.
- Record limitations and obtain the client’s acceptance before generalising results.
What we do not claim
RouteLens does not claim survey-grade GPS, frame-exact synchronisation, validated automatic defect detection, pavement measurements such as IRI or rut depth, legal admissibility, or guaranteed time savings. Physical, customer and field release gates remain unverified. Plan a pilot to measure suitability for your actual workflow.
Sources and further reading
Questions
Is the homepage demo a real customer survey?
No. It uses UGAC’s own GoPro test recording of a rural road, simplified to sampled frames and a thinned track without coordinates. It contains no customer footage, and the amber findings on it are illustrative examples, not survey evidence.
Does matching an extracted GPS track prove location accuracy?
No. Matching confirms extraction consistency. Field accuracy requires an independent reference and a stated method.
Can a pilot establish whether RouteLens fits our reporting workflow?
Yes. Agree sample footage, scope, acceptance criteria and deliverables, then record the measured outcome and any limits.
RouteLens is built by UGAC. Where this guide describes RouteLens behaviour, it reflects the release status published on 29 October 2026; see the product overview for what is verified.
