A genetic algorithm for rapid discovery of functional articulating magnetic microrobots

Microscale robotic devices promise advances in drug delivery, responsive materials, and compact energy storage, yet most rely on monolithic architectures that limit scalability and functional diversity. Realizing their potential requires predictive design informed by a deeper understanding of self-interactions, environmental dynamics, and field-driven responses. To address these challenges, we developed a modular fabrication approach in which magnetic subunits assemble into functional microrobots, greatly expanding design flexibility and scalability. However, the combinatorial design space—billions of possible configurations—necessitates predictive modeling for efficient exploration. We developed a Monte Carlo simulation framework capable of evaluating microrobot actuation behavior and quantifying metrics such as contraction, folding symmetry, and exerted force. We show that coupling this framework with a genetic algorithm enables directed evolution of virtual microrobot populations toward target behaviors, such as defined contraction ratios, prescribed folding pathways, or precise piconewton-scale force delivery. Populations evolve through simulated selection and point mutations applied to individual magnetic subunits, progressively refining structural performance. This evolutionary design paradigm establishes a physics-informed, data-driven route for discovering reconfigurable microrobotic systems with emergent, task-specific functionality.