Fluid turbulence is a longstanding challenge in physics and engineering. It can disrupt aircraft performance, reduce wind turbine efficiency, and contribute to circulatory problems when blood flow becomes irregular. Because turbulence is highly complex and difficult to predict, developing effective control methods has remained a major scientific challenge. Now, Lagemann and colleagues1 report the development of HydroGym, an artificial intelligence training platform designed to teach AI models how to control fluid flow through trial and error. The researchers demonstrate that control strategies learned in one fluid-flow environment can be transferred to more complex environments without additional training. If validated through experimental studies, HydroGym could advance turbulence control and have applications well beyond hydrodynamics, including aviation, renewable energy, and biomedical engineering.
Competing interests
The authors declare no competing interests.
Source: www.nature.com