Dynamic optimization of ant colonies for routing and recharging electric vehicles
We are delighted to present our new contribution to research into intelligent and sustainable logistics: “Dynamic Ant Colony Optimization for EV Routing & Charging”.
The challenge:
How can existing delivery routes be efficiently adapted to a fleet of electric vehicles, without rebuilding everything?
This is a major challenge for logistics operators in a context of energy transition.
Current delivery routes, designed for combustion engines, do not take into account :
– the limited autonomy of electric vehicles,
– the location of recharging stations,
– nor strict adherence to time slots.
Starting from scratch would be too costly. The objective is to intelligently reconvert existing routes to meet the new constraints.
Our solution:
In partnership with DEKI, we have developed EVROPT, a bio-inspired approach based on Ant Colony Optimization (ACO).
The idea: reproduce the collective behavior of ants to build, adapt and optimize routes.
Rather than generating new routes, EVROPT dynamically adapts existing thermal routes, inserting recharging stops in an optimized way.
Key innovations :
Dynamic pheromone deposition: taking into account charge level (SOC), distances to charging stations and time constraints.
Realistic energy model: consumption calculated according to vehicle physics (mass, gradient, friction, etc.).
Consolidated recharging network: aggregation and processing of open data to reflect a real network.
Strong, flexible constraints: strict adherence to time slots, penalties if battery falls below 20%.
Observed results:
The algorithm was tested on 7 real tours (Paris) initially planned for thermal vehicles.
Tours < 90 km: converted at 100%, without delay.
Tours > 90 km: partial conversion with recharge stops inserted.
Battery level always maintained > 20%.
Controlled lengthening: +39 to +65% time, +11 to +112% distance.
Practical applications:
– Support for thermal → electric fleet conversion
– Rapid EV feasibility simulation
– Reduced carbon impact without heavy reconfiguration
– Optimized segmentation of mixed fleets.