Conference Agenda

Efficiently managing supply chains is vital for modern defense and space operations to ensure security, rapid deployment, and resource maximization (Grala et al., 2024; Wang and Jin, 2020). This paper examines two critical planning layers: macro-level transport selection and micro-level cargo space optimization. Specifically, we investigate the integration of Data Envelopment Analysis (DEA)—a mathematical method used to calculate the relative efficiency of different options (Charnes et al., 1978; Lepchak and Voese, 2020)—and intelligent three-dimensional bin packing problems (3D-BPP), which involve placing items of different sizes into fixed containers (Lodi et al., 2002; Gonzalez-San-Martin, 2024). Crucially, we extend this model to the domain of space logistics, where extreme weight and volume limits, tight rocket launch windows, and massive transport costs amplify supply chain risks (Bacon et al., 2022; Grogan et al., 2018). Following a rigorous PRISMA-compliant screening protocol (Page et al., 2021) of peer-reviewed papers published between 2016 and 2026, we systematically map how multi-criteria efficiency tools interact with automated loading plans across earth and space-based environments.

Our systematic literature review (SLR) reveals a deep disconnect between these two planning layers. On one hand, DEA models are highly effective at establishing efficiency baselines to balance costs, orbital cargo risks, and strategic priorities when selecting transport modes (Chao and Yu, 2022). However, DEA outputs are rarely linked dynamically to physical loading constraints (Chu et al., 2025). On the other hand, computer-driven approaches to 3D-BPP—such as Deep Reinforcement Learning (DRL)—excel at generating tight loading layouts under strict safety regulations, including load stability, center-of-gravity calculations, and hazardous material separation (Gao et al., 2025; Kaleta, 2025). Yet, these packing algorithms lack the broad, multi-layered risk and strategic fleet evaluation inherent to DEA frameworks, particularly when navigating civil-military shared infrastructure under dynamic uncertainty (Pishvaee et al., 2011; Verma and Dynamic Group, 2023).

Based on our synthesis, we identify three critical research gaps in the current state-of-the-art:

1. A severe lack of real-time data integration under sudden operational and space-weather uncertainty (Ivanov et al., 2010; Szymczak, 2017).
2. Insufficient attention paid to shared civil-military fleet capacity allocation for space launch vehicles (Habibie, 2026).
3. A shortage of explainable AI (XAI) frameworks tailored to provide transparent justifications to human operators in safety-critical transportation environments (Gunning et al., 2019; Hu et al., 2026).

To bridge these gaps, we propose a new conceptual framework that feeds DEA efficiency scores directly as input states into reinforcement learning packing agents. This review establishes a comprehensive academic baseline for future empirical research, offering a structured, actionable pathway toward building autonomous, resilient, and eco-efficient earth and space-based defense supply chains (McKinnon, 2018; Sbihi and Eglese, 2010).