Simulation of fish locomotion and control

Abstract

This work discusses the development of a simulation of fish locomotion and its corresponding control. The simulation is based on unsteady panel methods and can calculate the loads on a general 2-D deformable hydrofoil (simplification of an aquatic animal) and then integrate them to produce the motion history. The code receives as input a specified body geometry history and outputs its loading and path. In order to achieve the reverse the simulation is used in conjunction with a neural network controller that predicts the needed body deflections (body geometry history) necessary to perform a specified maneuver. Validation results are presented that show the accuracy of the present simulation and several uncontrolled test cases are presented to show the capabilities of the controller.