Directed Acyclic Graph Model of AGVs Dispatching Charging and Optimal Scheduling

This paper presents a novel service graph model of task assignment and scheduling of Automated Guided Vehicles (AGVs) and its optimisation for smart manufacturing. The studied problem involves the loading sequences and collision avoidance at junctions of a multi-loop route network, complicating the model. In the proposed model, a material transport service in a workshop is formulated as a two-step process comprising a pickup-and-delivery (P&D) task and a backhaul task. A directed acyclic graph (DAG) model is developed to model the task assignment, where the P&D task is modelled as vertices and the backhaul task is modelled as the edges connecting a P&D task and its successive P&D tasks. In the backhaul task, the AGV select whether it will charge at the charging station. Charging and not charging are modelled as two edges (charging edge and non-charging edge) connecting two consecutive P&D nodes. Based on the DAG, a mixed-integer linear programming (MILP) model is developed to optimise the task sequencing and charging constraints for AGVs. Numerical experiments are conducted to evaluate the performance of the proposed AGV dispatching, scheduling and charging strategy.