Dynamic stochastic electric vehicle routing with safe reinforcement learning

Dynamic routing of electric commercial vehicles can be a challenging problem since besides the uncertainty energy consumption there are also random customer requests. This paper introduces Stochastic Electric Vehicle Routing Problem (DS-EVRP). A Safe Reinforcement Learning method is proposed for solving problem. The objective to minimize expected in safe way, which means minimizing risk battery depletion while en route by planning charging whenever necessary. key idea learn offline about stochastic requests and using Monte Carlo simulations, able plan predictively safely online. evaluated simulations based on data from realistic traffic model city Luxembourg high-fidelity vehicle model. results indicate that it possible save at same time maintaining reliability routes an anticipative way. has potential improve transport operations with capitalizing their environmental benefits.