# Could AI Spot Road Damage Before It Becomes a Pothole? Australian Trial Detects 300 Defects > Source: > Published: 2026-10-27 07:46:22+00:00 # Could AI Spot Road Damage Before It Becomes a Pothole? Australian Trial Detects 300 Defects ## RoughSense combines video, GPS and vehicle movement data to help authorities track deteriorating roads and prioritise repairs across Victoria An Australian trial has used [artificial intelligence (AI)](https://www.ibtimes.co.uk/uk-raf-ai-wildfire-action-plan-1827336) to detect early signs of road deterioration before they develop into potholes that can damage vehicles. RMIT University and the Royal Automobile Club of Victoria (RACV) fitted an AI system called RoughSense to an RACV Emergency Roadside Assistance van operating on Victoria's Mornington Peninsula. Between 21 August and 21 October 2026, RoughSense recorded more than 300 defects across 1,000 kilometres of Victorian roads. RoughSense combines AI, GPS tracking, video footage and vehicle movement measurements to identify road defects. RMIT researchers developed algorithms that merge images of road surfaces with movement and location data. ## AI System Combines Multiple Data Sources Instead of detecting only fully formed potholes, the system looks for early signs of deterioration, including surface cracks and rutting. Its findings are checked against defects identified manually by patrol driver Mark Orlando. The researchers are using these checks to calibrate the system and improve its accuracy and reliability throughout the trial. The van regularly travels the same routes, allowing RoughSense to monitor how individual defects change over time. Project co-investigators said repeated journeys had shown potholes worsening over time. Tracking these changes could help road authorities identify deteriorating surfaces before they require more extensive repairs. RoughSense identified a high concentration of defects along the Frankston–Flinders Road corridor, which covers Tyabb, Hastings, Bittern and Balnarring. The resulting maps provide councils and road authorities with location-specific information about deteriorating pavement. This could help them prioritise resurfacing and plan repairs rather than relying on ad hoc maintenance. Addressing minor defects early could also reduce long-term repair costs. ## Poor Visibility Increases Road Hazards Orlando said subtle cracks and shallow dips could be difficult for motorists to spot in wet weather, at night or on unfamiliar roads. Drivers may consequently hit hazards at speed. Automated sensors can record surface flaws without relying on a driver's ability to see them. Identifying and repairing defects before they become deep potholes could help prevent tyre blowouts, rim damage, loss of control and accidents. The project follows RACV's 2026 My Country Road survey, which collected 10,048 improvement requests from more than 6,000 regional Victorians. Potholes and crumbling surfaces were the main road safety concern for 71% of respondents, while 73% requested resurfacing. RACV combines survey feedback with Australian Road Assessment Program ratings and crash data to advocate for targeted investment in high-risk regional roads. ## Partners Prepare for Further Testing RMIT leads the research programme, system integration and development of the detection algorithms. Industry partner MetaSpatial contributes sensing hardware, engineering expertise, field testing, implementation and data visualisation. MetaSpatial director Dr Adam Chabok said combining industry expertise with university research helped turn technology into practical road safety tools. RACV provides the van and driver. The trial will continue collecting data into early 2027 to refine the system's data integration, reliability and ability to track defects across repeated journeys. Any wider rollout would require further coordination between researchers, industry partners, fleet operators and municipal road authorities. The next stage of an Australian road-monitoring project will investigate deploying the RoughSense artificial intelligence (AI) system across larger vehicle fleets. The aim is to extend continuous road monitoring to broader parts of Australia, according to RMIT. © Copyright IBTimes 2026. All rights reserved.