RoadAI – Intelligence to Drive
Move · Save · Win
Overview
RoadAI is a cutting-edge road condition monitoring system developed by Electronic Media Services. Designed to support smarter infrastructure maintenance, RoadAI uses AI-powered image analysis to detect and prioritize road damage—especially potholes—across urban environments.
Unlike traditional methods, RoadAI offers objective, consistent, and regularly updated insights, enabling local authorities to target repairs where they’re most needed.
How It Works
RoadAI is built around four integrated modules:
1. Capture Module
Mounted on vehicles that routinely cover local roads (e.g. refuse trucks), this module includes:
- A camera
- GPS
- Cellular modem
It captures images at regular intervals (up to 1 per second), sends scaled-down versions to the AI server, and—if potholes are detected—uploads full-resolution images and metadata to the database.
2. AI Analysis
The AI server:
- Detects potholes and categorizes them by size
- Identifies and blurs faces and license plates for privacy
- Optionally overlays insights directly onto images (e.g., location, severity)
3. Database Server
Stores all processed images and data securely. It ensures privacy compliance and supports scalable data management.
4. Visualisation Interface
Currently implemented on iPad for flexibility, with future options including:
- Web dashboards
- Windows applications
- Mobile apps
Users can:
- View pothole locations on a map
- Filter by type, date, region, and severity
- Click flags to see detailed data and full-size annotated images
Key Features
- Real-time detection of road defects
- Privacy protection via automatic blurring
- Severity classification for prioritization
- Interactive map interface for easy navigation
- Modular architecture for scalable deployment
Future Enhancements
- Integration with job management systems for automated repair scheduling
- Advanced overlays with contextual insights (e.g., lane position, hazard level)
- Expansion to detect other road hazards like puddles or cracks
AI Development
EMS trained custom deep learning models from scratch to meet the unique demands of road condition monitoring. The process included:
- Dataset curation: Selecting high-quality, relevant training data
- Model training: Tailoring algorithms for real-world accuracy
- Rigorous testing: Ensuring reliability across diverse conditions
- Server-side implementation: Efficient processing with minimal delay
Call to Action
Want to know more about how RoadAI can enable your local authority to target repairs where they’re most needed?