Navigation of a three-link microswimmer via deep reinforcement learning

Swimming microorganisms employ effective gaits to navigate toward specific targets. Equipping artificial microswimmers with similar capabilities presents significant challenges in motion planning and gait design. In this study, we explore the use of deep reinforcement learning to enable a three-link microswimmer to navigate using AI-advised swimming gaits. We highlight how the swimming gaits that emerge during the learning process depend on specific choices of the reward function. We also compare these results with optimal swimming gaits reported in previous studies.