Deadlock Detection and Avoidance Strategies for Automated Storage and Retrieval Systems

This paper focuses on real-time control of automated storage and retrieval systems (AS/RSs) serviced by a rail-guided vehicle system, a widely used solution for material handling in warehouses. The generic multiproduct AS/RS is modeled as a timed discrete event dynamical system, whose state provides the information on the current interactions between users and resources. Moreover, we address the real-time controller that governs resource allocations and scheduling choices by enabling and inhibiting the system events in order to avoid collisions and deadlocks. To this aim, we characterize deadlock in AS/RSs and define two deadlock resolution strategies: a deadlock avoidance and a deadlock detection/recovery policy. The proposed deadlock formulation and characterization have a general validity and can be applied to single unit resource allocation systems where a subset of users may be regarded as resources of other users. We compare the proposed control policies for a large-scale AS/RS presented in the related literature by several discrete event simulation tests.