SafeFuture Lab · rCITI@UNSW
SafeFuture Lab · rCITI, UNSW Sydney

Road-safety evidence that survives being questioned

We build statewide road-safety evidence from raw public data — signals, traffic volumes, road geometry, speeds and crashes — and every number we publish can be traced to its source record, re-run from raw data, and defended by the person who built the pipeline.

System integrity firstStatewide, open dataVerified AIMethods, limits and refusals published
At a glance

Scale we work at

4,527 / 4,588volumed signalised intersections with per-approach evidence (98.7%)
1,267,052Victorian road sections with directional traffic volume
34,220skewed intersections mapped against crash history
99spatial models in our QGIS library (2022–2026)
Five flagship proofs

What we have actually done

1

Statewide signal evidence

About 4,870 signal-configuration sheets — many scanned, with no text layer — read with AI vision and checked against road geometry, giving per-approach evidence at 4,527 of 4,588 volumed signalised intersections in Victoria.

2

Directional traffic volume for every road

1,267,052 Victorian road sections in three evidence tiers (measured, intersection, modelled), with about 71 billion vehicle-kilometres closed to a documented envelope — and the source disagreements published, not hidden.

3

Barry Road: we refused our own tool's refusal

Our before/after engine declined to report a small study. We rejected that: hiding a result is also a claim. It now reports CMF 0.75 (95% CI 0.30–1.85) — consistent with a reduction, not proof of one.

4

Speed-zone evaluation and aggregation bias

Six NSW speed-zone changes evaluated with weighted difference-in-differences, and a connected-vehicle method that separates path-level speeds from aggregate averages.

5

34,220 skewed intersections, open source

Every Victorian intersection classified by approach angle and linked to crash history and heavy-vehicle networks — public, with the code open.

Capabilities

Six areas of expertise

01

System integrity

Gated, reproducible pipelines; published limits; estimates with honest uncertainty.

02

Statewide network evidence

Signals, detector counts, directional volumes, camera counts and speed-limit truth.

03

Spatial road risk

Skew, vertical grade, curvature, Movement & Place and risk rating, joined to crashes.

04

Speed, evaluation & connected vehicles

Before/after methods, speed-zone studies and connected-vehicle validation.

05

Crash data systems & analysis

Crash reporting systems and national fatality analysis.

06

Verified AI & ML

Public AI agents that answer only what they can check; crash modelling.

How we work

Three commitments

Source before action.
Root cause before reply.
Every estimate ships with its uncertainty.
Live tools

Try them

Ask SafeFuture Lab
Road-safety questions answered only when the answer checks out against sources
Open →
Skewed intersections
34,220 Victorian intersections, angle bands and crash history
Open →
Road grade × crashes (3D)
Vertical grade profiles with crash points
Open →
Connected-vehicle speed audit
Probe speeds compared with instrumented research cars
Open →
AusRAP & network risk explained
Explainer with a live AI assistant
Open →