Speaker
Description
An autonomous boat with electric propulsion using a direct-current motors as prime movers and waterjet propulsors is investigated. Dynamical modeling capable to adequately describe the seakeeping of the craft in various conditions is employed to fuse fundamental principles and data sets obtained from an experimental campaign. A standard nonlinear state-space model is formulated to use as reference. Then, a Nonlinear Autoregressive Moving Average (NARMA) discrete-time model is developed and compared with the reference and the experimental results. Both models have neural networks in their core that are trained using supervised training machine learning methods. Both dynamical models form the basis for applying physicomimetic approaches to control and navigation of standalone boats or swarm thereof.
