Sensorimotor control of fish rheotaxis

Most aquatic animals can navigate in complex flow environments. Fish, for example, can consistently orient and swim against oncoming flow (rheotaxis) even without visual cues. Rheotaxis is a challenging process that involves multiple steps, including flow sensing and motor control. Previous studies proposed a data-driven model of rheotaxis in which fish rely on temporal changes in the local curl of the background flow, measured via the mechanosensory lateral line system. Here, we revisit this model in the context of a simplified mathematical fish, consisting of a self-propel dipole swimming in a two dimensional flow channel. We propose an ensemble of sensorimotor feedback laws, and we test these laws by carrying out stability analysis and parametric studies. Specifically, we compare two classes of control models, discontinuous time delayed models that stagger sensing and response and continuous models with instantaneous response. We find that both models can achieve rheotaxis but with distinct performanceand dynamic behaviors. These results suggest that the existence of many sensorimotor rules for rheotaxis and establish bio-inspired design rules for underwater robotics design.