U-M Aero Faculty Lecture: Ricardo Vinuesa on Zero-Shot Control and the HydroGym RL Environment

University of Michigan Associate Professor Ricardo Vinuesa explains how physics-guided surrogate learning makes it possible to train reinforcement-learning controllers in simple channel flows and deploy them, without additional training, on realistic turbulent wings. He demonstrates how this zero-shot approach can reduce training costs by four orders of magnitude while achieving significant reductions in skin-friction and total drag. Vinuesa also introduces HydroGym, an open-source reinforcement-learning platform for fluid dynamics that brings together benchmark flow-control environments, simulation tools, algorithms, and models. He discusses how shared resources and libraries of trained controllers could accelerate turbulent-flow research and support future numerical and wind-tunnel experiments.