Shellharbour City Council in New South Wales fitted council vehicles with AI-powered cameras that automatically detect and GPS-map potholes, cracks, faded line-marking, damaged signage and other defects across its 500km road and 300km pathway network, rating each defect by severi
500 km
Road network covered (2025)
300 km
Pathway network covered (2025)
Details
Maturity
Pilot
Promoter
Shellharbour City Council
Period
2025
Keywords
local government, public works, transportation, computer vision
Context
In October 2025, Shellharbour City Council (New South Wales, Australia) began operating a specialised AI camera mounted to the front of council vehicles to automatically scan roads and pathways as vehicles carry out their normal duties across the Shellharbour Local Government Area.
Objectives
The council's aim is to triage hazards faster and plan maintenance work better by logging defects automatically, rather than relying solely on resident-reported complaints.
Activities
The system identifies a wide range of defects - potholes, cracks, uneven surfaces, litter, overhanging vegetation, faded line-marking, damaged signage, graffiti and roadkill - and logs each with photographic evidence and a GPS location into a defect database, with every defect rated on a severity scale so maintenance crews can prioritise the most urgent repairs. Council plans to add a second camera before the end of 2025 to help survey its full 500-kilometre road network and 300 kilometres of pathways more quickly.
Results
Mayor Chris Homer described the system as showing 'great promise for increasing community safety and efficiency,' with the council citing faster hazard triage and better forward planning of maintenance work as the main expected benefits.
Conclusions
As a newly launched pilot, the council has not yet published before/after figures on repair-response times, backlog reduction or cost savings, and no detection-accuracy statistics (unlike some peer deployments) have been made public - this is an early-stage rollout to watch rather than a proven, evaluated outcome.
Implementation
Indicative cost
Low (< €50k) — One AI camera deployed October 2025, a second planned by end of 2025; cost not disclosed.
Time to results
Short (< 1 year) — Launched October 2025; still in early rollout at time of reporting.
Staffing & skills
Shellharbour City Council operational/maintenance staff
Conditions for success
Mounting AI cameras on council vehicles already doing normal duties, avoiding a dedicated survey fleet
Severity-rated defect database to prioritise maintenance crew response
Planned second camera (by end of 2025) to cover the full 500km road / 300km pathway network
Common failure modes
No detection-accuracy statistics published, unlike some peer deployments
No before/after repair-response, backlog or cost-saving figures yet, being only weeks into operation
Where it fits
Governance type
Australian local council
Scale
single local government area (500km roads, 300km pathways)
Income level
high income
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
San José mounted AI-powered cameras on city vehicles to automatically detect potholes and roadway debris, reporting 97% pothole-detection accuracy and …
MLIT runs a national catalog certifying AI and sensor-based bridge/tunnel inspection technologies against published performance criteria. Listings grew from 12 …