Exploring the Interactions Between a Free Swimmer and Near-Wall Flows Using Deep Reinforcement Learning

The capability to navigate in the proximity of solid surfaces while avoiding collision and maintaining high efficiency is essential to the design of underwater vehicles. In this work, the hydrodynamics of swimming in close proximity to a solid surface is explored. Near-wall interactions of neutrally buoyant, or negatively buoyant animals are similar to ground effects experienced by fixed-winged aircrafts near a horizontal surface. Therefore, the free swimmer can experience changes to lift and drag during locomotion. Reduced drag can benefit the swimmer, however, changes in lift may lead to a collision with obstacles. Additionally, swimming close to walls can reduce the range of undulatory motion, thereby limiting performance. To study these effects in detail, the current work uses deep reinforcement learning coupled with two-dimensional numerical simulations of self-propelled swimmers. The artificial swimmers utilize mechanosensory inputs similar to the lateral line in biological fish, which detect pressure and velocity in the surrounding flow. The swimmers are trained to autonomously discover optimal actions, which allow them to navigate reliably and efficiently in the presence of obstacles. The behavior exhibited will be examined to explore how the interactions between a free swimmer and near-wall flow fields affect propulsion, efficiency, and kinematics.