Conversational AI for Rapid Scientific Prototyping: A Case Study on ESA's ELOPE Competition

TL;DR

The paper investigates whether conversational AI can accelerate scientific prototyping by analyzing a case study in which ChatGPT contributed to code, algorithm design, and data handling for ESA's ELOPE competition. The approach demonstrates that an LLM-assisted team can achieve competitive results (second place with a score of $0.01282$) while offering concrete guidance such as fixed-number-event windowing, but also exposes limitations including bloated code, detours in dialogue, and occasional silent errors, necessitating careful human oversight. It offers a structured set of best practices—branching discussions, version control, test-driven development, lean prompts, stepwise construction, and visualization—to enable reliable AI-assisted prototyping in science. The work argues for deliberate integration of LLMs into scientific workflows to speed up prototyping while maintaining rigor, acknowledging that AI capabilities are rapidly evolving and that these recommendations may need updating over time.

Abstract

Large language models (LLMs) are increasingly used as coding partners, yet their role in accelerating scientific discovery remains underexplored. This paper presents a case study of using ChatGPT for rapid prototyping in ESA's ELOPE (Event-based Lunar OPtical flow Egomotion estimation) competition. The competition required participants to process event camera data to estimate lunar lander trajectories. Despite joining late, we achieved second place with a score of 0.01282, highlighting the potential of human-AI collaboration in competitive scientific settings. ChatGPT contributed not only executable code but also algorithmic reasoning, data handling routines, and methodological suggestions, such as using fixed number of events instead of fixed time spans for windowing. At the same time, we observed limitations: the model often introduced unnecessary structural changes, gets confused by intermediate discussions about alternative ideas, occasionally produced critical errors and forgets important aspects in longer scientific discussions. By analyzing these strengths and shortcomings, we show how conversational AI can both accelerate development and support conceptual insight in scientific research. We argue that structured integration of LLMs into the scientific workflow can enhance rapid prototyping by proposing best practices for AI-assisted scientific work.

Figures (9)

• Figure 1: Start of co-development of algorithm for ELOPE challenge.
• Figure 2: Comments of ChatGPT about proposed approach (see Fig.\ref{['fig:chat-init']}).
• Figure 3: Prompt for asking to implement event agglomeration into images.
• Figure 4: Prompt for asking to discuss the implementation of a homography estimation from the images.
• Figure 5: First response to the request for discussing a homography implementation (see Fig. \ref{['fig:chat-homopgraphy']}).