This thesis investigates the development of a flexible and interactive design configurator using NVIDIA Omniverse, with a focus on modular design integration, usability, and simulation validation. The study explores the platform's capabilities in building customizable 3D configurations through Python scripting and the Universal Scene Description (USD) format. Two distinct configuration systems were developed: one for train components and another for fixtures. These systems demonstrate the adaptability of a modular codebase and user interface design within Omniverse.

The research further evaluates Omniverse’s physics simulation tools by implementing a three-point bending test. While the visual representation of simulations is effective, the platform falls short in delivering scientifically accurate results. These findings suggest that Omniverse is more appropriate for visual prototyping and scene assembly than for engineering-level analysis or detailed design automation.

Despite certain limitations, such as the lack of parametric modeling and precise assembly tools, the project concludes that Omniverse is well-suited for complex, modular visualization tasks and collaborative design environments. The thesis delivers not only configurators but also a structured development workflow and a manual to support future users in similar applications.