AGENT: An Aerial Vehicle Generation and Design Tool Using Large Language Models

TL;DR

This work presents AGENT, a UAV design assistant built on the CodeT5+ LLM and trained directly on the AircraftVerse JSON dataset to perform design generation and surrogate evaluation. By organizing a domain-specific curriculum (component generation, design generation, design masking, and surrogate evaluation), AGENT learns to generate designs conditioned on flight-dynamics properties and to predict simulator outcomes, effectively acting as a surrogate for the physics model. With only 220M parameters, AGENT achieves strong surrogate performance and feasible conditional generation while running on a single GPU, enabling rapid, iterative design workflows. The results demonstrate significant potential for deploying domain-specific LLMs in co-design pipelines, accelerating exploration and reducing reliance on expensive simulations, with concrete guidance on runtime considerations and human-in-the-loop integration.

Abstract

Computer-aided design (CAD) is a promising application area for emerging artificial intelligence methods. Traditional workflows for cyberphysical systems create detailed digital models which can be evaluated by physics simulators in order to narrow the search space before creating physical prototypes. A major bottleneck of this approach is that the simulators are often computationally expensive and slow. Recent advancements in AI methods offer the possibility to accelerate these pipelines. We use the recently released AircraftVerse dataset, which is especially suited for developing and evaluating large language models for designs. AircraftVerse contains a diverse set of UAV designs represented via textual design trees together with detailed physics simulation results. Following the recent success of large language models (LLMs), we propose AGENT (Aircraft GENeraTor). AGENT is a comprehensive design tool built on the CodeT5+ LLM which learns powerful representations of aircraft textual designs directly from JSON files. We develop a curriculum of training tasks which imbues a single model with a suite of useful features. AGENT is able to generate designs conditioned on properties of flight dynamics (hover time, maximum speed, etc.). Additionally, AGENT can issue evaluations of designs allowing it to act as a surrogate model of the physics simulation that underlies the AircraftVerse dataset. We present a series of experiments which demonstrate our system's abilities. We are able to achieve strong performance using the smallest member of the CodeT5+ family (220M parameters). This allows for a flexible and powerful system which can be executed on a single GPU enabling a clear path toward future deployment.

Figures (3)

• Figure 1: Graphical representation of a UAV JSON design tree from the AircraftVerse dataset.
• Figure 2: Curriculum of training tasks.
• Figure 3: Example generation using multiple prompts. We ran the generated design into the physics simulator to verify the design. In practice, this would be optional.