Introduction
Swindon Borough Council's award-winning Emerging Technology team acts as an agile innovation unit dedicated to testing and scaling modern technologies — including Generative AI, data science, and Robotic Process Automation (RPA) — across public services. Using a "fail or win fast" framework, the team rapidly deploys practical technologies to improve the efficiency of council operations, lower costs, and enhance the lives of local residents.
The Challenge
Across England and Wales, local councils face a massive infrastructure challenge: research from the Asphalt Industry Alliance (AIA) shows that only 51% of local road networks are in good condition, with national pothole repairs estimated at £18.6 billion. Making matters worse, traditional maintenance is inherently reactive — relying on residents to discover and report issues like potholes, graffiti, or fly-tipping via online forms. Swindon Borough Council set out to break this cycle by leveraging data science and AI to transform public service delivery from reactive fixes to proactive prevention.
What We Did
Our founder, Jillur Quddus, in collaboration with Swindon Borough Council's Emerging Technology team, designed and built an award-winning AI system, and accompanying Android mobile application, capable of autonomously detecting, classifying and prioritising cases of potholes and fly-tipping in real time via live video feeds. Powered by a custom-designed and trained deep convolutional neural architecture, the on-device system processes videos at 45fps with under 25ms latency — delivering up to 40 detection events per second — and integrates seamlessly with live GPS location services.
In the video below, an Android phone mounted to the dashboard of a Swindon Borough Council street-cleaning vehicle demonstrates our system in action. Every on-device detection event automatically captures the pothole image, bounding box data, and exact GPS coordinates. This data is then transmitted to a processing server for deduplication and integration with the council's existing cloud-based case management system. Should connectivity drop, the app caches data locally and automatically syncs once a connection is re-established.
Key Outcomes
Directly embedding exact GPS data into the council's case management system enabled street-cleaning crews to locate potholes much faster than deciphering free-text reports submitted by members of the public. Recognising the system's impact, Swindon adapted the AI model to detect instances of fly-tipping — delivering an 83% boost in operational efficiency by eliminating labour-intensive manual processes.