The Agentic Aggregator Framework

This research introduces an agentic framework designed to manage the complex, multi-variable environment of electric bus fleet operations. The system couples a traditional optimization-based scheduling model with a supervisory agentic layer. This layer is responsible for detecting operational disturbances (such as route delays or energy deviations), adapting to electricity tariff changes, and evaluating schedule feasibility in real-time.

By integrating these agents, the system can trigger re-optimization selectively, ensuring that the fleet maintains service reliability while maximizing the use of charging infrastructure and Vehicle-to-Grid (V2G) opportunities. The agentic layer acts as a bridge between the physical constraints of the fleet (battery state-of-charge, charger availability) and the economic incentives of the grid.

Operational Trade-offs and Policy Risks

The study evaluates the framework through a realistic depot case study, comparing profit-based versus operation-based coordination modes. While agentic aggregation successfully maintains feasible schedules and improves flexibility, it introduces a significant economic trade-off.

When the agentic system is configured to prioritize profit, it can effectively extract value from the Public Transport Operator (PTO). This creates a tension between the efficiency of the aggregator and the operational costs of the transit provider. The authors conclude that for agentic systems to be viable in public-fleet contexts, they must be deployed with:

• Transparent coordination modes that clearly define objectives.
• Auditable tariff-setting mechanisms to prevent predatory pricing.
• Explicit, pre-defined value-sharing rules to ensure equitable distribution of benefits between the aggregator and the PTO.