RoadAI

RoadAI originated on a provincial road network audit. The client required objective, repeatable defect classification across 180+ kilometres of road in a timeframe that fitted a single budget cycle. Foot-based manual audits had historically taken four to six months and produced inconsistent grading.

Off-the-shelf pavement management systems either required dedicated survey vehicles the authority did not own, or produced output that could not be aligned with TMH9 without significant manual rework.

We built RoadAI to close that gap. Proprietary AI defect detection tuned to TMH9 categories, GPS-accurate positioning integrated with wheel-speed distance sensors, and an output format that dropped directly into the authority’s existing asset management system.

RoadAI detects, classifies, and locates road defects from vehicle-mounted camera footage captured at normal traffic speeds. The classification is deliberately aligned to TMH9 pavement condition categories so the output can be tendered against without translation work.

The pipeline is optimised for high-throughput processing, making network-scale audits tractable within days rather than weeks. ASTM D6433 export is also available for international engagements.

AI does not have the final word. Detections are reviewed by the responsible engineer, and the audit report is signed by a registered Pr Eng before it is issued. The model accelerates the engineering, it does not replace the judgement or the accountability behind it.

• Proprietary AI defect detection model tuned on South African and international road datasets
• Distance-based frame sampling to ensure no double-counting and no missed defects at highway speeds
• GPS and wheel-speed integration for accurate defect positioning
• TMH9 and ASTM D6433 compliant output formats
• Map-based DeltaCloud dashboard for client review and prioritisation
• CSV export for direct import into client asset management systems

RoadAI is deployed on road infrastructure engagements: provincial and national road authorities, municipal networks, industrial site roads, and mine haul roads.

The tool is particularly effective for annual or quarterly re-audit cycles, where the objective, repeatable classification provides a meaningful trend line for asset condition over time.