Methodology for Automating Multi-Robot Baggage Handling in Airports

As small and medium-sized airports face growing logistical pressures, the need for autonomous, scalable, and energy-efficient baggage handling systems has become increasingly urgent. This study presents a reinforcement learning–based framework for coordinating multiple mobile robots in a simulated airport environment characterized by partial observability, constrained pathways, and energy limitations. In addition, this paper evaluates and compares four reinforcement learning algorithms—Proximal Policy Optimization (PPO), Deep Q-Network (DQN), Deep Deterministic Policy Gradient (DDPG), and Q-Learning—on their ability to manage task sequencing, collision avoidance, and energy conservation. The custom-built simulation incorporates spatial congestion, state-of-charge (SOC) monitoring, and dynamic mission planning across multiple agents. Results demonstrate that PPO significantly outperforms other algorithms in success rate, energy efficiency, and navigation stability, underscoring its potential for real-world deployment in complex, multi-agent environments. This work contributes a replicable simulation platform, a comparative performance evaluation across algorithms, and a coordination strategy adaptable to decentralized robotic systems operating under real-world constraints.