AI Helps Electric Delivery Vans Charge Smarter, Cutting Long Waits

Running a fleet of electric delivery vehicles is tricky. Drivers often need to recharge mid-shift, but if too many vans arrive at the same station, they end up waiting in long queues. One obvious fix would be a central computer that assigns each van a charging time and place. But most real-world charging networks don't support that kind of control.

Instead, researchers from Spain took a different approach. They gave every van a learning AI system that acts independently. Each van looks at its own delivery deadlines and battery level, plus a live feed of how crowded nearby charging stations are. Then it makes its own choice about where to go and how much power to take. No central command, no messages between vans — just each one making a smart local decision that leads to smooth overall coordination.

In computer simulations using real street maps from 20 cities, these AI-driven vans completed more than 98% of deliveries on time, even in cities they'd never seen during training. A simpler rule — just driving to the nearest station when the battery runs low — only managed 73%. The AIs also learned clever habits like taking short, partial charges instead of always topping off, and avoiding stations that looked too busy. As a result, average charging wait times plunged from roughly 45 minutes to under 2 minutes.

The breakthrough isn't about the technology itself — it's that this near-perfect coordination emerges without expensive, centralized systems. For companies testing electric delivery fleets, this could mean lower operating costs and more reliable service. It also hints at a future where AI handles complex logistics with minimal infrastructure, making it easier for businesses to go electric.