An Adaptive BDI Agent-Based System Design for Distributed 3D-Container Loading

Efficient loading systems are crucial in contemporary e-commerce and manufacturing, as they determine the delivery time of products. The spatial allocation of products within containers and time scheduling for loading tasks within loading areas are both NP-hard problems, where computational complexities grow exponentially with system input size. Existing algorithms designed for loading issues, such as greedy and genetic methods, are hard to cope with uncertainties and lack adaptability in distributed environments, due to the centralized decision-making structure. This paper addresses both the threedimensional container loading problem (3D-CLP) and loading task scheduling to benefit real-world loading problems. An agentbased loading system is proposed, which leverages decentralized BDI (Belief-Desire-Intention) agents to extend loading algorithms from centralized to decentralized structures. The proposed system has advantages in reducing computational complexity, enabling intelligent decision-making in open environments, and offering scalability through a plug-and-play design.