String Grammars for Preliminary UAV Design Exploration

Unmanned Aerial Vehicles (UAVs) have proven useful for a substantial number of tasks in areas such as transportation and surveillance. However, the design process of creating UAVs is an incredibly complex task that lies at the intersection of multiple disciplines such as aerospace engineering and fluid dynamics. Given this, exploring the incredibly large design space of all possible UAVs for performant designs is cost-prohibitive, time-prohibitive and creatively draining. Designing novel UAVs manually is slow and also risks the loss of time and effort wasted in the event of a middling or sub-par performance. Moreover, it may not always be clear how to improve a design that is generated manually but performs inadequately with respect to simulation performance, aerodynamic posture, or material cost. In this paper, we present a method to create preliminary structural designs for UAVs by leveraging a string-based grammar and accompanying parser to quickly and efficaciously produce feasible design geometries which can be improved further by a human designer or through downstream automated filtering. We evaluate this approach on Computer-Aided Design (CAD) components that were designed by subject matter experts in aerospace engineering. We demonstrate the efficacy of our approach by providing vehicle designs assembled in PTC Creo.