New AI-Era Design Framework Generates 864 Aircraft Concepts to Pick the Best Air Taxi

The skies above the world’s rapidly growing cities are on the verge of a transportation revolution, but engineering the aircraft that will fill them has turned out to be far harder than anyone predicted. Advanced air mobility, or AAM, vehicles are unlike anything in the existing aviation catalog. They rely on distributed electric propulsion, they switch between hover and wing-borne cruise in a single flight, and they must operate safely in congested urban airspace where noise, footprint and safety margins are unforgiving. The result is a design space so vast and so unconventional that the traditional, intuition-driven methods of early aircraft design are struggling to keep up. A new study published in the International Journal of Aeronautical and Space Sciences offers a systematic answer, and it could reshape how the next generation of flying machines is born.

Researchers Min Ji Kim, Hyeon Jun Lee and Jae-Woo Lee of Konkuk University in Seoul have unveiled an advanced knowledge-based design framework that transforms the earliest and most consequential phase of aircraft development, the generation and selection of concepts. In conventional practice, a handful of experienced designers sketch out a few candidate layouts, debate their merits, and converge on a favorite through judgment shaped by years of experience. That process has produced the aircraft we fly today, but it carries two structural weaknesses. It explores only a tiny corner of the possible design space, and it depends heavily on the individual designer, which makes outcomes inconsistent and hard to reproduce. When the design space includes hundreds of possible combinations of lift systems, propulsion architectures and fuselage arrangements, human intuition alone cannot map the territory.

The Korean team’s framework attacks the problem on both fronts by integrating five interlocking components: functional decomposition, a knowledge-based database system, systematic morphological analysis, logical concept family grouping, and formal filtering and scoring procedures. Functional decomposition breaks the aircraft down into what it must do rather than what it should look like, generating a neutral list of functions such as lift generation, propulsion, energy storage and passenger accommodation. Each function is then paired with a menu of candidate solution options drawn from a structured knowledge base, ensuring that the concept generation step begins from evidence rather than from habit.

The heart of the framework is the knowledge-based database system, or KBDS, which the team implemented using Obsidian, a networked note-taking platform that allows engineers to link technologies, products and competitors into a navigable web of design knowledge. To populate the database for their case study, the researchers surveyed 67 regional air mobility, or RAM, competitor aircraft, catalogued 102 distinct technologies and compiled 182 product entries. Each entry captures the performance characteristics, maturity level and configuration role of a real-world solution, giving the design team a searchable library of what has already been attempted and how well it works. The use of technology readiness levels adds a practical filter, flagging solutions that are promising on paper but too immature for a near-term aircraft.

With the requirements fixed and the knowledge base assembled, the framework turns to morphological analysis, a method with a distinguished pedigree. First applied to astronomy by Fritz Zwicky in the late 1940s, morphological analysis arranges design problems as a matrix of configuration components and their possible solutions, then exhaustively combines them to enumerate the full solution space. For the RAM case study, the combination of key configuration components and their solution options produced 864 candidate aircraft concepts. That number is precisely the point: no conventional brainstorming session would ever surface 864 distinct, technically grounded alternatives, and many of the most promising ones might never cross a design team’s whiteboard.

Generating hundreds of concepts, however, creates a new problem, namely how to organize and prune them without reintroducing the subjective bias the framework was designed to eliminate. The Korean team’s answer is a statistical clustering technique they call logical concept family grouping. The method measures the configurational similarity of every candidate concept, essentially how closely each one resembles the others in the population across its chosen solutions. Using statistics borrowed from the behavioral sciences, including Cohen’s d effect sizes and contrast ratios, the framework quantifies which solution choices genuinely distinguish one group of concepts from another and assigns each component-solution pairing a family grouping score. The 864 candidates were thereby organized into eight coherent concept families, each representing a distinct and internally consistent design philosophy rather than an arbitrary slice of the list.

Filtering then reduced the field to half of the generated candidates by screening out configurations that violated constraints or duplicated dominant solutions without offering meaningful advantages. The surviving concepts faced a final, multi-criteria evaluation in which figures of merit were weighted using the analytic hierarchy process, a structured pairwise-comparison method that aggregates the judgments of multiple evaluators into a consistent set of priorities. By combining weighted figures of merit with the family structure established earlier, the team narrowed the field to a single final favorable concept for the regional air mobility mission, a result that emerged from a documented, repeatable chain of reasoning rather than from the preferences of any single engineer.

The significance of the work extends beyond one aircraft study. The framework explicitly incorporates AAM-specific features, such as distributed electric propulsion, multi-mode flight capability and urban operational constraints, into the generation process itself, so that unconventional configurations are treated as first-class candidates rather than exotic outliers. Because every step is procedural, from the functional decomposition to the statistical grouping to the weighted scoring, two independent teams applying the same framework to the same requirements should arrive at the same shortlist. That consistency and reproducibility are exactly what regulators, manufacturers and investors need as the AAM industry matures from bold demonstrations into certified, revenue-carrying services.

The timing could hardly be better. Dozens of companies worldwide are racing toward entry into service for electric vertical take-off and landing aircraft, and early design choices, made years before flight test, lock in much of an aircraft’s eventual cost, noise footprint, safety record and certification path. A framework that widens the search while disciplining the selection promises to reduce the risk of betting an entire program on a configuration that a broader exploration would have exposed as inferior. The methodology also aligns naturally with modern knowledge management tools, suggesting a workflow in which an engineering organization’s cumulative experience, captured in a living knowledge base, continuously feeds and improves its concept generation engine.

The researchers acknowledge that the framework does not remove human judgment entirely; evaluators still assign the weights that reflect program priorities, and the knowledge base reflects the scope of the survey behind it. But by converting intuition-heavy steps into measurable, documented procedures, the Konkuk team has moved aircraft conceptual design closer to an engineering science and further from an art. As advanced air mobility fleets prepare to transform how people move through cities, the ideas that take flight first may increasingly come not from a single designer’s sketchbook, but from an exhaustive, statistically disciplined search across 864 futures.

Subject of Research: A knowledge-based design framework for aircraft concept generation and selection in advanced air mobility

Article Title: An Advanced Knowledge-Based Design Framework for Aircraft Concept Generation and Selection

Article References: Kim, M. J., Lee, H. J., & Lee, J.-W. (2026). An Advanced Knowledge-Based Design Framework for Aircraft Concept Generation and Selection. International Journal of Aeronautical and Space Sciences. https://doi.org/10.1007/s42405-026-01296-3

Image Credits: AI Generated

DOI: 10.1007/s42405-026-01296-3

Keywords: advanced air mobility, aircraft design, knowledge-based system, concept generation and selection, design framework, distributed electric propulsion, regional air mobility, morphological analysis, multi-criteria decision-making, eVTOL, Advanced, Knowledge-Based