This work presents a structured pipeline for co-creative drone light show design that integrates large language models (LLMs) with analytical validation and deployment systems. Rather than treating LLMs as autonomous choreographers, the study reframes them as semantic front ends that generate interpretable creative intent, which is then translated into physically valid and executable drone formations. The proposed framework couples language-based generative interfaces with analytic solvers and the Skybrush Studio API, which performs safety verification, trajectory optimization, and compilation into deployable show formats. Through this integration, we demonstrate how semantic creativity and syntactic rigor can coexist within a single workflow, enabling intuitive yet verifiable design of drone swarm performances. The findings clarify where language-driven generation adds value, where it requires analytic reinforcement, and how modular architectures can support collaboration between human designers, generative models, and production-grade control software.
Watch the team present this project at 01:25:12 in the session recording here.
Keywords: drone light shows, large language models, semantic-syntactic integration, Skybrush Studio API, generative design pipeline, creative robotics, human-AI collaboration, formation sampling, trajectory optimization
Faculty Advisor
Steve Barry serves as technology and data leader at a global investment bank. He brings more than 25 years of leadership experience in the realms of technology and finance. Furthermore, Steve’s expertise extends to data science and AI, with a demonstrated proficiency in platforms like AWS, Azure, and GCP.
Steve is currently a Lecturer in the Master’s in Applied Data Science at the University of Chicago. He graduated with a master’s degree from the same program in 2020. His background in finance and technology, has been pivotal in applying data science in practical contexts.
Complementing his professional experience, Steve served as a Teaching Assistant in the same University of Chicago Master’s program for three years for multiple courses including: BDP, DEPA, ML, and Data Visualization. This role allowed him to bridge the gap between academic knowledge and its application in the finance and tech industries. Steve is an enthusiastic lifelong learner that provides him unique perspective to communicate complex concepts clearly.