Digital Twin of an Intersection for Intelligent Traffic Management and Safety

Clients faced difficulties in optimizing traffic and improving road safety—especially for vulnerable participants such as pedestrians and cyclists. The limitations of existing sensors (primarily detection loops) prevented detailed analysis and rapid responses. This resulted in an increased risk of accidents, inefficient traffic management, and a lack of data insights on traffic both in real time and in the form of statistics.

The solution (Yutraffic awareAI) is a local road AI system that creates a real-time digital replica of the traffic situation. It uses advanced deep-learning video analysis to detect, classify, and track road users via multiple cameras. It offers applications such as intersection hotspot analysis, traffic stream analysis, or response applications tailored to user needs. The solution is remotely accessible via a web browser and fully configurable with an installation wizard, simplifying management and automating traffic control.

• Service engineers and project managers – signing acceptance protocols and providing technical support
  
• Administrators and operators – system configuration, application management, and performance monitoring
  
• Technicians – hardware installation, diagnostics, and maintenance
  
• Traffic engineers – analysis of road user behavior and data-driven decision-making
  
• Universities and Smart City projects – development of custom applications using raw data

Product development and testing: 1.5 – 3 years
Configuration and delivery to customers: 14 days

• K3S (Kubernetes) for orchestration
  
• Java (backend applications)
  
• React (frontend applications)
  
• Python (system-level scripts)
  
• Keycloak for identity and access management
  
• Grafana for monitoring
  
• MQTT and Mosquitto (for communication with third-party systems)
  
• AWS services for backup and remote management
  
• Weblate for application localization
  
• Playwright for automated testing

Head of business unit

Full implementation

• Internal team (service)
  
• Customer

• Knowledge of the monitored and managed intersection for effective use of the straightforward web application, guided by the supplied manual or training.

• IT infrastructure when connecting to the customer’s network

• Detailed insights into road user movements in real time → more effective traffic flow optimization
  
• Increased safety, especially for vulnerable users (pedestrians, cyclists)
  
• Support for autonomous driving through extended contextual awareness of traffic situations
  
• Data-driven decision-making for traffic engineers → higher quality decisions
  
• Enhanced data protection (anonymization zones)
  
• Lower operational costs and effort thanks to remote updates
  
• Simpler user experience with intuitive interface and installation wizard

Advanced AI video analysis enabling comprehensive detection and tracking of road users without blind spots, including less common types (e-scooters), contributed to the project’s success. Smooth integration with traffic infrastructure (controllers) and central systems was also crucial. Another key factor was that the product was designed and implemented as a general solution, able to cover a wide range of customer problems—every customer is different, every situation is different, but the system remains the same. This universality is essential for mass deployment.

The diversity of intersections where the system is installed → possibility for configuration and adjustment by the customer using a well-designed graphical interface, usable even without a manual.

• Prioritize data protection – anonymize and have DPIAs ready for tenders
  
• Invest in integration – ensure smooth communication with existing and future infrastructure. Customers will want data for their own analyses
  
• Use remote management – monitor, alert, and update the system remotely to cut costs and ensure availability
  
• Empower users – give operators tools to configure independently, reducing reliance on specialized support