Training microrobots to swim by a large language model

Abstract

Machine learning and artificial intelligence have recently become a popular paradigm for designing and optimizing robotic systems across various scales. Recent studies have showcased the innovative application of large language models (LLMs) in industrial control and in directing legged walking robots. Here, we use an LLM, Generative Pre-trained Transformer 4 (GPT-4), in most scenarios, to train two prototypical microrobots for swimming in viscous fluids. Adopting a few-shot learning approach, we develop a minimal, unified prompt composed of only five sentences. The same concise prompt successfully guides two distinct articulated microrobots---the three-link swimmer and the three-sphere swimmer---in mastering their signature strokes. These strokes, initially conceptualized by physicists, are now effectively interpreted and applied by the LLM, enabling the microrobots to circumvent the physical constraints inherent to microlocomotion. In addition, we extend our study to the four-link and four-sphere swimmers, enabling their effective propulsion. Remarkably, for this specific task, our LLM-based decision-making strategy substantially outperforms a traditional reinforcement-learning method in terms of training speed. We discuss the subtle aspects of prompt design, particularly emphasizing the reduction of costs when using GPT-4.