ELECTRIC VEHICLE ROUTING PROBLEM USING DEEP REINFORCEMENT LEARNING

Abstract

The Vehicle Routing Problem (VRP) is a well-known combinatorial optimization problem with numerous practical applications, such as optimizing delivery routes, transportation logistics, and mobile resource allocation. Traditional methods for solving VRP often rely on heuristics and mathematical programming techniques, which may struggle to handle large-scale instances and dynamic environments. In recent years, reinforcement learning (RL) has emerged as a promising approach for addressing complex optimization problems. In this thesis, we propose a novel framework for solving the VRP using reinforce ment learning techniques. Our approach leverages the power of deep RL algorithms, specifically the combination of deep neural networks and actor critic, to learn ef fective policies for route planning and optimization. By formulating the VRP as a Markov Decision Process (MDP), we develop an RL agent that learns to make sequential decisions on vehicle movements and load allocations. We evaluate our proposed RL framework on a set of benchmark VRP instances, comparing its performance against traditional heuristics and optimization techniques. The experimental results demonstrate that our approach achieves competitive so lution quality and computational efficiency, especially in larger problem instances. Furthermore, we investigate the robustness and generalization capability of the learned policies by evaluating their performance on unseen problem instances. Overall, our work highlights the potential of reinforcement learning as a promis ing methodology for solving the Vehicle Routing Problem. Bycombiningthestrengths of deep RL algorithms and problem-specific insights, we show that RL can offer ef ficient and effective solutions to this challenging optimization problem, paving the way for further advancements in the field of transportation logistics and route plan ning.