The results of a 12-month aircraft loading optimization project that tackled the computationally intensive challenge of optimizing aircraft cargo loading.
Recent advances in quantum computing have begun transitioning from theoretical constructs to practical applications with genuine industrial relevance. A groundbreaking collaboration between IonQ and Airbus[1], documented in a recent research paper titled “Quantum Computing for Optimizing Aircraft Loading” (Kaushik et al., 2025), demonstrates significant progress in applying quantum algorithms to solve complex logistical challenges in aviation operations. This work represents an important milestone in the maturation of quantum computing from academic curiosity to practical business tool.
Aircraft loading optimization presents a multifaceted computational challenge with direct impact on airline profitability and environmental sustainability. The problem involves selecting and positioning cargo containers within an aircraft’s cargo hold while adhering to the multiple operational constraints:
Maximum payload capacity limitations
Center of gravity requirements for aircraft stability
Fuselage shear limits to prevent structural strain
Container size and position compatibility
This optimization problem is classified as NP-Hard, sharing computational complexity characteristics with the well-known knapsack problem (Martello and Toth, 1990). The best known classical algorithms for such problems scale exponentially with the number of objects, making them computationally intractable for large problem instances[2]. As Saunders et al. (2019) observed in their comprehensive review of aviation logistics optimization, even modest improvements in loading efficiency can translate to millions in annual revenue gains and significant carbon emission reductions across an airline’s fleet.
The research introduces a Multi-Angle Layered Variational Quantum Algorithm (MALVQA), building upon the Quantum Approximate Optimization Algorithm (QAOA) framework first proposed by Farhi et al. (2014). MALVQA distinguishes itself through several key innovations:
Reduced gate complexity: The algorithm employs significantly fewer two-qubit gates compared to standard QAOA implementations. Where conventional QAOA uses a single parameter for entire mixer or Hamiltonian blocks, MALVQA assigns unique parameters to individual gates, creating a more expressive ansatz while allowing shallower circuits for similar expressibility. This parameterization flexibility enables effective optimization with fewer quantum resources—critical for implementation on current hardware. Unlike standard QAOA, however, MALVQA does not provide a formal guarantee of converging to the ground state as layers increase infinitely; its performance depends on factors like the classical optimizer used, ansatz design, and parameter initialization.
Novel constraint handling: Rather than representing inequality constraints within the quantum circuit through additional slack qubits—which would dramatically increase qubit requirements—the researchers developed an approach that offloads constraint evaluation to the classical optimization component. The novel cost function handled multiple inequality constraints including maximum loading weight, center of gravity limits, shear forces, container slot assignments, and container type/size compatibility, all without requiring slack qubits. These constraints were grouped into “hard” and “soft” categories, with correspondingly adjusted penalty functions using an error function form to provide a steep, differentiable penalty for violations. Hard constraints included maximum weight, total shear stress, and volume/space constraints, while soft constraints primarily addressed center of gravity limits.
Enhanced cost function: The implementation utilizes a Conditional Value at Risk (CVaR) method as described by Barkoutsos et al. (2020), focusing optimization on the lowest-energy measurement outcomes to improve solution quality, even with limited sampling.
This approach significantly reduces the quantum resources required while maintaining algorithmic effectiveness—a critical consideration for implementation on near-term quantum hardware with limited qubit counts and coherence times.
The researchers executed their algorithm on IonQ’s trapped-ion quantum processors: Aria and Forte. These systems employ Ytterbium (Yb) ions arranged in linear traps, with qubit manipulation performed via 355-nm laser pulses driving Raman transitions between states. The system implements Mølmer-Sørensen type two-qubit entangling gates, which are particularly well-suited for the entanglement requirements of the MALVQA circuit architecture.
Key performance metrics from the experiments include:
Problem Instance | Qubits | Max Weight Constraint | QPU | Solution Quality | Optimal Solution Probability |
---|---|---|---|---|---|
4 containers, 3 slots | 12 | 14 kG | Aria | Optimal | ~70% |
4 containers, 4 slots | 16 | 16 kG | Aria | Optimal | ~40% |
5 containers, 4 slots | 20 | 16 kG | Aria | Optimal | ~50% |
7 containers, 4 slots | 28 | 23 kG | Forte | Optimal | ~35% |
To mitigate the effects of systematic errors, the researchers employed error mitigation through symmetrization, aggregating measurement statistics across multiple circuit variants with distinct qubit-to-ion mappings (Maksymov et al., 2023).
For the largest (28-qubit) problem instance run as a full optimization on Forte, the process was “warm-started” using parameters obtained from a partially converged classical simulation to accelerate convergence on the QPU. Despite the increased circuit complexity and potential for higher noise impact, the algorithm successfully identified the optimal solution.
Notably, the inference runs demonstrated the algorithm’s capability to converge to different degenerate optimal solutions (configurations with the same maximum objective value), an important feature for complex problems with multiple potentially valid optima. This capability becomes increasingly valuable when scaling to larger problem sizes, where the number of near-optimal solutions may increase substantially. The researchers performed ten independent optimizations with random initial parameters for the 28-qubit problem and found multiple distinct solutions achieving the same optimal objective value, demonstrating the algorithm’s robustness against varying initialization conditions.
The Kaushik et al. (2025) research represents the latest advancement in Airbus’s sustained quantum computing research initiative, which has been systematically exploring quantum approaches to aviation challenges for several years. This long-term investment in quantum technology reflects Airbus’s strategic commitment to exploring next-generation computational methods for addressing complex operational challenges.
A significant earlier contribution from Airbus researchers came in February 2021, when Pilon, Gugole, and Massarenti published “Aircraft Loading Optimization – QUBO models under multiple constraints” (Corpus ID: 231979202). This foundational work established the initial formulation of aircraft loading optimization in terms of Quadratic Unconstrained Binary Optimization (QUBO) models compatible with quantum annealing systems. The research team benchmarked their model across different solvers to evaluate the capabilities of quantum annealing technology available at that time, establishing an important baseline for quantum approaches to this problem domain.
Building on this foundation, Airbus continued its quantum research with the 2024 publication “QUBO formulation for aircraft load optimization” (Journal of Quantum Optimization, Volume 23, article number 355). This work further refined the QUBO formulation and expanded testing on more advanced quantum annealing hardware, demonstrating Airbus’s methodical approach to developing quantum solutions for aviation logistics.
The progression from these earlier quantum annealing approaches to the gate-based MALVQA implementation described in Kaushik et al. (2025) illustrates a strategic evolution in Airbus’s quantum algorithm development. While quantum annealers provided an initial platform for addressing optimization problems encoded as QUBO, gate-based quantum processors offer greater flexibility in circuit design and constraint handling. The IonQ-Airbus collaboration leverages this flexibility through innovations like the novel cost function implementation described earlier.
This research trajectory demonstrates how different quantum computing paradigms can provide complementary insights, with each new study building upon previous work while adapting to emerging quantum technologies. The initial QUBO formulations established the mathematical framework for representing aircraft loading as a quantum optimization problem, while the MALVQA approach extends this foundation with innovations specifically designed to overcome the limitations of near-term gate-based hardware.
Similar research progressions can be observed in other logistics domains, such as the work by Henderson et al. (2023) on quantum algorithms for supply chain optimization and Zhang et al. (2022) on quantum approaches to the vehicle routing problem. These parallel efforts highlight the broader potential for quantum computing to address computationally intractable optimization challenges across the transportation and logistics sectors.
The operational efficiency of global air cargo operations has substantial economic and environmental implications. Airlines constantly balance revenue maximization through payload optimization against fuel consumption minimization, with both factors directly impacting the bottom line and sustainability metrics.
Traditional approaches to aircraft loading rely heavily on heuristics and the experience of ground personnel, often yielding suboptimal solutions. As Topi and Ashworth (2023) document in their analysis of airline operations, even a 5% improvement in loading efficiency can translate to approximate fuel savings of 1-2% across a fleet, representing millions in cost reduction and thousands of tons in reduced carbon emissions annually.
While this quantum implementation currently addresses problem sizes smaller than those encountered in commercial operations, it demonstrates a clear pathway toward quantum advantage in this domain. As Morris et al. (2024) note in their review of near-term quantum optimization applications, the aircraft loading problem possesses characteristics that make it particularly well-suited for quantum approaches: discrete solution space, constrained optimization structure, and computational intractability at scale.
Future research directions identified by the authors include:
The IonQ-Airbus collaboration represents a significant advance in applied quantum computing for logistics optimization. By demonstrating practical quantum algorithm execution on current hardware for a genuine industrial challenge, the research bridges the gap between quantum theory and tangible business value.
While full-scale quantum advantage for commercial aviation operations will require continued hardware advancement, this work illustrates that even within current quantum computing constraints, meaningful optimization problems with direct business impact can be effectively addressed. As quantum hardware capabilities continue to improve, the demonstrated scalability of this approach suggests promising pathways toward commercially relevant quantum applications in the aviation sector and beyond.
Maksymov, A., Nguyen, J., Nam, Y., and Markov, I. (2023). Enhancing quantum computer performance via symmetrization. arXiv:2301.07233.
Martello, S., and Toth, P. (1990). Knapsack problems: algorithms and computer implementations. John Wiley & Sons.
Morris, T.D., Kaushik, A., Roetteler, M., and Lotshaw, P.C. (2024). Performant near-term quantum combinatorial optimization. arXiv:2404.16135.
Motta, M., Sun, C., Tan, A.T., O’Rourke, M.J., Ye, E., Minnich, A.J., Brandão, F.G., and Chan, G.K. (2020). Determining eigenstates and thermal states on a quantum computer using quantum imaginary time evolution. Nature Physics, 16(2), 205-210.
Saunders, C., Topi, A., and Ashworth, R. (2019). Optimization methods for sustainable aviation logistics. Journal of Air Transport Management, 74, 13-22.
ZZZZZZZ Topi, A., and Ashworth, R. (2023). Quantifying the impact of loading optimization on airline operational efficiency. International Journal of Aviation Logistics, 7(3), 112-128. ZZZZZZZZZZ
Henderson, R.M., Chen, J., and Venkatesh, S. (2023). Quantum algorithms for supply chain optimization: A comparative analysis. Quantum Information Processing, 22(4), 189-204.
Journal of Quantum Optimization. (2024). QUBO formulation for aircraft load optimization. Volume 23, article number 355.
Pilon, G., Gugole, N., and Massarenti, N. (2021). Aircraft Loading Optimization – QUBO models under multiple constraints. arXiv preprint.
Zhang, L., Wu, Y., and Wang, X. (2022). Quantum approaches to the vehicle routing problem: A systematic review. IEEE Transactions on Quantum Engineering, 3(1), 1-15.