ScholarWorks@Y-Scholar Hub

초록

Automated guided vehicles (AGVs) are responsible for transporting materials within a warehouse. They require continuous task allocation, must perform assigned tasks without collisions and deadlocks, and need to maintain adequate battery charge to improve throughput while preventing depletion. Designing a suitable control architecture and optimising the model for AGV systems in dynamic environments poses an intricate challenge due to the computational demands associated with each flow. We introduce a model that employs a multiagent system to manage the AGV system, incorporating both task assignment and battery charging procedures. Viewed from an algorithmic standpoint, the task allocation procedure mitigates collision and deadlock risks, while the battery charging procedure ascertains the appropriate charging station and timing for each visit. In this model, both task allocation and battery charging operations adopt a market-based approach among multiple agents, enabling each agent to make optimal decisions based on its available data. Consequently, the computational and decision-making burdens are distributed among individual agents. Experimental results indicate that the proposed model is competitive with existing benchmark models, outperforming them by at least 6.7% in dynamic environments with frequent task occurrences. The proposed model holds promise for contributing to the advancement of AGV applications in warehouse management.