Title: AI-Based Sorting Strategies for Automation in Logistics and Supply Chains
Authors: Nesreen S. Aljerjawi and Samy S. Abu-Naser
Volume: 9
Issue: 8
Pages: 94-98
Publication Date: 2025/08/28
Abstract:
The explosive growth of e-commerce has led to unprecedented challenges in logistics and supply chain management, particularly in the domain of package sorting. Traditional sorting systems, often relying on fixed rule-based or simple algorithmic approaches with an average time complexity of O(nlogn), are increasingly inadequate for handling the high volume, velocity, and variability of modern operations. This paper presents a novel framework for an intelligent, adaptive sorting system powered by Reinforcement Learning (RL). The proposed system employs a Deep Q-Network (DQN) agent that learns to optimize sorting decisions in real-time based on a dynamic state space, including package attributes, conveyor belt congestion, and destination priorities. The model was trained and evaluated within a custom-built simulation environment. Comparative analysis against a baseline rule-based greedy algorithm demonstrated significant performance improvements. The RL-based system achieved a 21.15% reduction in average sorting time, a 68% decrease in the error rate, and a 23.47% increase in throughput. The findings highlight the immense potential of integrating AI, specifically Reinforcement Learning, to create robust, self-optimizing sorting systems that are essential for achieving full automation and efficiency in future supply chain ecosystems.