Efficiency Optimization and Three Dimensional Flow Visualization of Flapping Flight

Finding optimal locomotion strategies for small fliers or swimmers poses a serious challenge due to the complex fluid behaviors in conjunction with the vast kinematic parameter space available. Recent advancement of machine learning techniques, especially the policy search methods, can be potentially used to solve such problems. In this work, an evolutionary-strategy-based policy search algorithm is applied to a robotic wing to optimize flapping-wing trajectories that maximize the average lift generated per power consumption. The robotic wing have two degrees of freedom, i.e. stroke and pitch rotation which are transcribed with periodic triangular and trapezoidal functions, respectively. The learning experiments are repeated for four different prescribed stroke amplitudes while the Reynolds Number (Re) is maintained at 1000. The efficiency is observed to increase with the stroke amplitude and the lift is mainly generated through delayed stall. Moreover, advanced wing rotation is the preferred strategy for lower stroke amplitude. We also perform Particle Tracking Velocimetry (PTV) measurement of the flow created by the optimized trajectories to evaluate the underlying physics of efficient lift generation.