Multi-Agent Pickup and Delivery in Transformable Production

This paper focuses on multi-agent pickup and delivery (MAPD) in the context of the transformable production system. With the increasing demand for personalized products in today's globalized world, manufacturers are providing customers with the option to purchase items with specific details. Transformable production is a solution to manufacture customized products, which involves both static agents responsible for manufacturing and mobile agents responsible for pickup and delivery tasks. Numerous works in the literature have studied the MAPD problem. However, these works consider pickup-and-delivery tasks to be independent of each other. In the transformable production system, a delivery task can only start if the corresponding manufacturing process is finished. Therefore, it is necessary to consider the task dependencies when addressing MAPD in transformable production. To tackle this challenge, we propose a framework to solve this problem. We employ a heuristic algorithm to assign tasks to static and mobile agents and introduce a parameter that adjusts the expected pickup and delivery duration. Additionally, we introduce heuristic cost functions for path finding tailored to our problem. Through comparative analysis using synthetic test sets, we highlight the significance of the parameter used in pickup-and-delivery task scheduling in obtaining improved results. Furthermore, we demonstrate that our proposed algorithm and cost functions achieve slightly better performance than the baseline algorithm.