Construction of a virtual simulation experimental platform for unmanned aerial vehicle-based logistics

Against the backdrop of rapid smart city development, unmanned aerial vehicle (UAV) logistics has become an indispensable component of modern logistics systems due to its unique advantages of high flexibility, low cost, and high efficiency. This places higher demands on talent cultivation in the field of logistics engineering and management. However, traditional teaching methods are constrained by physical space and hardware costs and can struggle to effectively demonstrate the complete process that must integrate UAV technology, path planning, and scheduling optimization. This results in a significant disconnect between the teaching content and actual industry applications. Simulation technology has emerged as a pivotal tool for innovation in engineering education reform by leveraging its dual strengths of visualizing theoretical knowledge and enabling interactive operational workflows. Therefore, grounded in new engineering education concepts and the advanced philosophy of deep industry–education integration, this work establishes and implements an integrated simulation teaching platform for UAV-based logistics. The platform integrates key components, including UAV selection, mathematical modeling, optimization, and evaluation.

Based on the operations research optimization theoretical framework and intelligent algorithms, this work systematically constructs the mathematical model, intelligent algorithm, and simulation environment for UAV-based logistics. Its three core principles are: (1) The system model. Based on the classical vehicle routing problem and its variants, we establish a mixed-integer programming model that comprehensively considers factors such as the quantity of goods delivered, the UAV type, the complexities of terrain, and airspace constraints. The model minimizes the total transportation distance cost and the total number of UAVs required, and further extends the scheduling optimization model of multimachine cooperative operations. (2) The algorithm solution. To address the complex NP-hard problems, various metaheuristic methods, including genetic algorithm, simulated annealing, and whale optimization algorithm, were used to solve the model and optimize the parameters. We introduce a deep reinforcement learning algorithm to achieve the innovative adaptive and intelligent decision-making ability of the system in uncertain environments. (3) Simulation verification. We built a virtual environment for UAV distribution with the integrated simulation engine, which can visually run the distribution schemes generated by a variety of algorithms (such as calculating the key performance indicators, including task completion time, total cost, or UAV utilization rate). Additionally, the simulation experimental platform for UAV-based logistics is composed of the following five core functional modules, forming complete closed-loop teaching: a system cognition module (for theoretical learning and scene familiarity), scheme design module (for model construction and algorithm selection), simulation debugging module (for parameter adjustment and process observation), scheme verification module (for the operational scheme and data collection), and comprehensive evaluation module (for multidimensional performance analysis and experimental report generation).

The platform has three contributions to UAV knowledge: First, it establishes a mathematical model of path planning and scheduling optimization for UAV distribution, which provides a theoretical basis for quantitative analysis. Second, the platform not only integrates the traditional optimization algorithm but also introduces a reinforcement learning algorithm to effectively solve the complex problem model. Moreover, the platform has a friendly human–computer interaction interface that supports the whole process of visual control and real-time interaction, which improves the user experience and teaching effect. Additionally, a typical teaching case is introduced based on real goods delivery scenarios in mountainous areas. The results show that the UAV distribution scheme optimized based on this platform is more efficient than traditional vehicle distributions, which verifies the huge advantages of UAV logistics in special scenarios.

Through systematic simulation experiments on UAV-based goods delivery, this work enables students to thoroughly analyze the fundamental principles, core functions, and operational workflows of the delivery simulation system. It models the entire process of drone delivery system optimization—from cognition, analysis, debugging, and validation to evaluation. The platform also produces integrated, multistage, and multicategory adaptable simulations. This research provides critical technical support and a reusable teaching case for the deep application and promotion of simulation technology in intelligent logistics and holds significant reference value for advancing teaching model innovation and constructing high-level experimental platforms in logistics engineering, management science and engineering, and related disciplines at higher education institutions. This work has far-reaching implications for implementing new engineering education concepts and cultivating interdisciplinary talents equipped with systems thinking, innovation, and practicality.