The transition of the Low-Altitude Economy (LAE) from vehicle development to operational scaling raises a fundamental methodological question: under which conditions can established methods from industrial intralogistics be transferred to aerial logistics operations without violating aviation-specific physical, safety, and regulatory constraints. This perspective paper proposes a conceptual and methodological framework for the systematic convergence of industrial intralogistics and LAE operations, treating the LAE as a high-frequency material flow system. We consider this a promising research avenue that we are beginning to pursue systematically. Drawing on expertise from the Technical University of Munich (TUM) and Shanghai Jiao Tong University (SJTU), we identify topological analogies between automated 3D-shuttle warehouses and urban air corridors, and propose Material Flow Control (MFC) as a complement to classical Air Traffic Management (ATM). The core contribution is the Bounded Adaptation framework: a structured, dimension-based transfer model determining which intralogistics methods are portable to LAE, which require bounded modification, and which are inadmissible. The proposed methodology couples an MFC dispatching layer (upper level, Discrete-Event Simulation) with continuous flight-dynamics simulation (lower level), supplemented by Rare Event Simulation and Specific Operations Risk Assessment (SORA) 2.5-aligned Ground Impact Footprint assessment as the enabling safety complement. In this perspective, we present ten transfer parallels grounded in validated projects of the Chair of Materials Handling, Material Flow, and Logistics (fml), mapped to LAE challenges. The primary European contribution to Chinese LAE development lies in Safety Mindset and structured assessment methods—a productive research complementarity: China’s GB 46750-2025 1 s position reports provide precisely the data stream that European Rare Event Simulation requires for calibration. The overarching research question—which operational capacity can be expected when SORA 2.5 requirements are fully met—frames the joint TUM×SJTU research agenda that this paper derives. Quantitative validation will be reported in subsequent publications.
Accurate prediction of airfoil aerodynamic coefficients is essential for aircraft design. In this study, we propose a lightweight dual-path con-volutional neural network to predict airfoil aerodynamic coefficients. Unlike single-input networks that rely on either a geometric descrip-tion or a set of flow parameters, the proposed architecture processes both through two parallel paths: a convolutional path that encodes 64×64 airfoil shape images, and a fully connected path that encodes the Reynolds number and angle of attack. The extracted features are merged and passed to two regression heads, so that CL and CD are predicted by separate output branches rather than by a single shared head. Training data combine UIUC wind-tunnel measurements with XFOIL results, yielding 12,304 samples with Reynolds numbers from 3.92×104 to 5.14×105 and angles of attack from − 10.33∘ to 20∘. The complete model contains 289,870 parameters and was trained in 71 min on a standard workstation; unseen configurations are then evaluated in approximately 3 ms per case—several orders of magnitude faster than conventional CFD. The trained network achieves a coefficient of determination of 0.940 for CL with a mean absolute percentage error of approximately 0.86
With the widespread application of unmanned aerial vehicles in wireless communications, their open channels pose a dual challenge of concealment and security. The study focuses on intelligent reflection-assisted unmanned aerial vehicle communication systems, aiming to achieve the concealment of the communication process and the security of data transmission, and constructs a joint communication optimization model including unmanned aerial vehicles, users, eavesdroppers, and intelligent reflective surfaces. Under a hybrid optimization framework that satisfies strict concealment constraints, relaxation variables are introduced to represent the minimum average safety capacity of users, and the solution is carried out using the minimum secrecy rate optimization algorithm of deep double-Q networks and sequential convex approximations to achieve end-to-end collaborative optimization. Experimental results show that the proposed method achieves a security capacity of 2.41 Mbps while satisfying the concealment constraints. The total energy consumption of the research method is only 4.82 kJ, significantly lower than all other algorithms. The average energy efficiency of the unmanned aerial vehicle is as high as 5.94 Mbit/J, which can effectively meet the requirements of covert communication and secure transmission. The framework proposed in the study has significant advantages and feasibility in balancing dual security goals and building a highly reliable unmanned aerial vehiclecommunication network.
To address the problem of target loss caused by visual servo lag in unmanned vehicle dynamic target tracking, this study proposes an unmanned vehicle trajectory generation model that integrates target detection and trajectory prediction into the feedforward channel of visual servoing. This model uses the YOLOv8 algorithm and long short-term memory neural network for dynamic target tracking and trajectory prediction, and then employs a nonlinear image-based visual servoing model technology to generate control instructions, achieving real-time control and feedback of the trajectory generation process. Experiments based on simulations and real UAV platforms show that the proposed model can better eliminate the interference of obstacles, with an obstacle avoidance rate of 99.02
To address the capacity allocation problem of the energy system for long-endurance operation of a stratospheric airship, a comprehensive energy system model is developed, including solar radiation, photovoltaic (PV) power generation, propulsion and payload power consumption, and battery energy storage. Based on this model, multiple energy management strategies are considered, and the energy balance constraints are formulated. The feasible combinations of PV area and battery capacity are systematically analyzed, and the feasibility boundaries under different strategies are extracted to investigate the constraints and their variation characteristics in capacity allocation. The results show that there exists a clear trade-off between PV area and battery capacity. An increase in PV area reduces the required battery capacity; however, the marginal benefit decreases as the PV scale grows. The energy management strategies significantly affect the shape of the feasibility boundaries. The Maneuverable wind resistance and Position Energy Storage (PES) strategies exhibit steep boundary slopes in the region of small PV areas, along with noticeable curvature variations or inflection characteristics over certain ranges. In contrast, the Position Potential (PP) and fixed-speed strategies present smoother boundaries with an approximately linear trend. Furthermore, seasonal and latitudinal conditions have a significant impact on energy availability. Under higher latitudes and winter conditions, reduced solar radiation leads to a notable increase in battery capacity requirements, whereas spring, summer, and autumn conditions at mid-latitudes provide a broader feasible capacity region.
