Optimized City Buses
Industry
Transport
Client
Vilnius City Municipality
Team
BUILD
Year
2023
We were hired to deliver a turn-key AI solution to make public transport run on time – and enforce the rules – without us hiring a whole new IT department.
Challenge
Vilnius’s bus network suffered from inefficient scheduling and traffic rule violations – commuters waited too long (buses bunching or idling) and frequent bus-lane intrusions slowed service. Traditional traffic systems lacked real-time analytics to fix these issues.
Approach
Installed AI cameras (stationary & on buses) to detect incidents like illegal bus lane usage and to count passenger loads. Used computer vision models to flag violations and overcrowding in real time.
Leveraged 5G connectivity for instantaneous data upload from moving buses to a central system. Implemented a streaming data pipeline for live alerts to traffic controllers.
Developed an ML forecasting model for ridership, using historical and real-time data to predict passenger volumes and adjust bus dispatch frequencies dynamically.
Deployed a predictive maintenance module analyzing telematics (engine, braking data) to predict bus breakdowns before they happen.
Architecture/Backend
5G-connected AI cameras, data pipeline, telematics ingestion, ridership forecasting.
AI/ML models
Computer-vision models, ML ridership forecasting, predictive-maintenance modeling.
Infrastructure/Deployment
5G uplink from buses to the central platform; city-wide AI camera enforcement.
Key results
Average bus waiting times reduced by ~20% on pilot routes, thanks to real-time route optimization and dispatching. Commuters see faster arrivals and less overcrowding during peak hours.
Bus lane violations decreased ~15% after city-wide camera enforcement powered by AI, directly improving traffic flow and fairness.
Breakdowns down ~10% due to proactive maintenance alerts, increasing fleet availability. Overall, these efficiency gains cut fuel costs and emissions (fewer idle or redundant trips).
