Extreme-edge computing is an emerging paradigm that utilizes the computational capabilities of end devices to perform localized data processing, thus enabling responsive and adaptive operations in industrial settings. This article addresses the challenge of dynamic scheduling and workload distribution in multiautomated guided vehicle (AGV) systems, where tasks include both movement (e.g., navigation, loading and unloading) and data processing (e.g., onboard computation). To capture realistic conditions, a dynamic service request model is adopted, in which each request consists of a set of movement and/or data processing operations. We first present a mathematical formulation to formally define the constraints associated with resource assignment and movement operations. Building on this, we propose a novel heuristic, queue-aware scheduling and deadlock mitigation strategy (QASDMS), for assigning operations to AGVs. QASDMS considers factors, such as AGV locations, resource availability, execution queues, and potential deadlocks, while supporting the arrival of dynamic unpredictable requests. The objective is to minimize the total time required to complete the operations described in the service requests. To evaluate system performance, the resource intensity index (RII) is introduced, which measures CPU and RAM usage relative to the number of completed operations. Simulation results in both medium and large-scale scenarios show that QASDMS improves the total number of completed operations by over 190% compared to a baseline case, while keeping CPU and RAM usage below 21% . Furthermore, QASDMS achieves significantly lower RII values (over 60% reduction), indicating more efficient resource utilization. These results highlight the potential of QASDMS to improve performance in multi-AGV extreme-edge environments.
Except where otherwise noted, this item's license is described as Atribución 4.0 Internacional