This research includes the development of a computational model created to autonomously navigate a biomimetic underwater vehicle. The navigation procedure uses as input a set of vehicle geometric and state variables and from that it predicts the needed vehicle body deformations so that the vehicle can navigate through a set of given waypoints. The first task is to develop a two dimensional numerical simulation based on an unsteady panel method coupled with the vehicle dynamics. Numerous test cases spanning the range of possible conditions are processed with the current simulation and this data is subsequently used to train a neural network. The trained network can predict what body deflection time history is necessary for the vehicle to navigate a given set of waypoints. Validation results are presented that show the accuracy of the present flow solver and several test cases of autonomous navigation are presented to show the capabilities of the current method.
展开▼
