Fast adaptive control of a non-linear system by an ADALINE: motion in a fluid

We report results from our endeavour to build control systems for a small autonomous underwater vehicle (AUV), named C-SCOUT.We prove that an ADALINE with higher order inputs can learn the non-linear dynamics of a vehicle moving in a fluid. We then let one serve as the system model in an adaptive control system for the vertical motion of a small submarine. The resulting control system possesses all the features one could desire: it learns rapidly, then performs flawlessly, and is able to regain control of the vehicle if incidents modify its behavior. The technique we describe can straightforwardly be generalized to many other situations.