A DQN-Based Cargo Loading Task Planning Method for Multiple Service Desks

Addressing the cargo loading task planning problem in environments with multiple service desks of heterogeneous efficiency, this paper proposes an intelligent optimization method based on Deep Q-Networks(DQN). Traditional logistics scheduling methods face challenges in high-dimensional state representation and local optima traps when handling the strong coupling relationships among service desk efficiency differences, cargo compatibility constraints, and dynamic task allocation. This study quantifies efficiency disparities by constructing a service desk-cargo efficiency matrix, designs a composite state space integrating remaining cargo quantities and service desk timelines, and innovatively introduces an adaptive reward function that combines time cost, load balancing, and constraint penalties. Experimental results demonstrate that the proposed method reduces the total completion time of parallel operations across multiple service desks by 8.2% compared to traditional heuristic algorithms, while adhering to transport vehicle capacity and cargo compatibility constraints. This approach provides a novel theoretical framework and technical implementation pathway for dynamic task scheduling in complex logistics scenarios.