A team of young engineers and a university student in Thessaloniki has developed an award-winning AI-driven system to transform how electric buses operate in the city. The project, called SKGRID, was created during the Thessaloniki Public Transport 2.0 Hackathon, organised by the city's transport authority (OSETH).
The team consists of two software engineers, Leonidas Boutsikaris, 29, and Giorgos Katrilakas, 32, along with 20-year-old mechanical engineering student Filippos Athanasiadis. The three shared first prize with another group, addressing the challenge of designing an "action plan for a zero-emission bus fleet".
Their solution reimagines Thessaloniki as a 1,083-cell spatial grid, each representing a cluster of bus stops. Using a combination of spatial graph neural networks and temporal modelling, SKGRID can predict traffic congestion, energy consumption, charging demand and route timing in real time. This enables smarter decision-making for routing and charging electric buses, reducing unnecessary mileage, improving energy efficiency and enhancing overall system performance.
Unlike existing planning tools, which assume ideal operating conditions, the team built their model around the reality of constantly changing urban environments. Factoring in variables such as traffic congestion, weather conditions and even topographical features like steep roads, the system dynamically proposes optimal routes. This could help drivers avoid unnecessary depot charging trips, saving both time and energy.
The idea was partly inspired by an earlier tool developed by CERTH for Thessaloniki's bus operator, which focused on scheduling and charging plans but did not fully account for real-world variability. Building on available data and insights from the hackathon, the team introduced a more flexible, adaptive approach.
"It was a wonderful experience to work on urban mobility and sustainable transport for Thessaloniki and to discuss ideas with researchers, engineers and experts in the field," Boutsikaris said. Beyond the competition, the team is now aiming to further develop SKGRID, with ambitions to establish it as a startup.
By Maria Mathiopoulou - adapted from Greek by Vassia Barba
