Dynamic stochastic electric vehicle routing with safe reinforcement learning

Dynamic routing of electric commercial vehicles can be a challenging problem since besidesthe uncertainty of energy consumption there are also random customer requests. This paperintroduces the Dynamic Stochastic Electric Vehicle Routing Problem (DS-EVRP). A Safe Rein-forcement Learning method is proposed for solving the problem. The objective is to minimizeexpected energy consumption in a safe way, which means also minimizing the risk of batterydepletion while en route by planning charging whenever necessary. The key idea is to learnoffline about the stochastic customer requests and energy consumption using Monte Carlosimulations, to be able to plan the route predictively and safely online. The method is evaluatedusing simulations based on energy consumption data from a realistic traffic model for the cityof Luxembourg and a high-fidelity vehicle model. The results indicate that it is possible tosave energy at the same time maintaining reliability by planning the routes and charging in ananticipative way. The proposed method has the potential to improve transport operations withelectric commercial vehicles capitalizing on their environmental benefits.