Artificial microswimmers via reinforcement learning

There is a growing interest in developing artificial microswimmers that can self-propel like swimming microorganisms for potential biomedical applications. A fundamental challenge is the design of effective locomotory gaits that can overcome the constraints due to the dominance of viscous forces. In addition, the swimmer needs to adapt its gaits in order to re-orient and reach the target locations. In this talk, we report our progress in integrating machine learning techniques in the design of artificial microswimmers. We will discuss how reinforcement learning can be leveraged to enable complex maneuvers of the swimmer. The results demonstrate the vast potential of this new approach in designing smart microswimmers that can perform sophisticated tasks.