A COMPETITIVE DRL APPROACH FOR LAST MILE DELIVERY WITH A ROBOT FLEET

In recent years, the optimisation of delivery systems in urban environments has gained increasing interest, particularly due to advancements in Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL) techniques. This thesis explores the application of RL algorithms in a multi-agent simulation environment, with the goal of improving the efficiency of parcel deliveries. Using SUMO, a traffic simulation platform, along with Python and TRACI, a model was developed in which autonomous vehicles compete to optimise the delivery process. The thesis focuses on the competitive approach, examining in detail the effect of competition between agents on delivery time and operational costs. The selected approach has a significant impact on delivery efficiency and can represent a promising solution to optimise autonomous transport systems. This study provides insights for future developments in the field of intelligent logistics and autonomous fleet management.