Fish-Inspired Navigation via Flow Sensing in an Autonomous Robotic Swimmer

Autonomous ocean-exploring robots promise to greatly enhance the rate at which we can explore ocean environments. However, such robots must overcome the challenge of navigating through unknown ocean currents and seeking out areas of interest without prior knowledge of their surroundings. Inspired by the ability of aquatic animals to navigate via flow sensing, we constructed a palm-sized robotic swimmer as a test platform for flow-based navigation strategies. The robot is equipped with eight pressure sensors to mimic the function of canal neuromasts found in the lateral lines of fishes. As an analogy for tracking hydrothermal vent plumes in the ocean, the robot is tasked with locating a turbulent jet flow in a 6 by 6 by 16-foot water tank. A deep reinforcement learning algorithm runs onboard a high-speed microcontroller, which trains a neural network to steer the robot towards the jet plume using its real-time pressure measurements. Learned navigation strategies are interpreted in the context of the flow physics of the turbulent plume to better understand the working principles and potential limits of flow-based navigation.