Cluster scheduling method for automated guided vehicles based on an improved deep Q-network

An improved deep Q-network (DQN) algorithm enhanced with a self-attention mechanism is proposed for manufacturing workshop applications. A mathematical model for auto-guided vehicle (AGV) cluster scheduling is formulated, incorporating the self-attention mechanism into the DQN framework via the Transformer encoder structure. The model is solved using a combination of a workshop data model and a Markov model. Experimental results across various task scales demonstrate the effectiveness and robustness of the algorithm.