Adaptive signal processing
| Hi-index | 0.00 |
We demonstrate that simple adalines can learn to mimic the complex non-linear dynamics of a thruster propelled vehicle that moves in a fluid. This is made possible by our supplying them with judiciously chosen non-linear inputs. We then use such adalines to provide the internal model on which we base an adaptive motion controller for such a vehicle. The latter thus becomes able of intelligent behavior, in that it can predict ahead of time what its dynamical state will be, and can adjust its reactions accordingly. It will also be able to adapt to the incidents that modify its behavior or its environment. We demonstrate that our control system performs flawlessly and, in particular, is able to keep the vehicle perfectly still, without it developing the limit cycle observed in most marine vehicles. The well known ability of adalines to adapt rapidly endows this controller with all the features that can be wished for in a neural network system. It learns by itself, and it learns so rapidly that it can be given the responsibility of the vehicle with adaline weights initialized to zero. The fact that simple adalines suffice for this complex task is important in itself because the technique we describe can straightforwardly be generalized to many other situations.