Neural networks for truck backer-upper control system

Artificial neural networks have been studied for many years in the hope of achieving human-like performance in many fields. One of these fields is to use neural networks to solve highly nonlinear control systems. The multilayer feedforward neural network has proven to be a very powerful tool in this field. Nevertheless, such a network suffers from a very time-consuming training procedure. In this paper, based on radial-basis function and recurrent neural networks, the authors develop two different neural network systems for backing up a computer simulated truck to a loading dock in a planar parking lot. In the authors' simulations, these two neural networks are capable of learning from the training samples and performing generalization, and therefore provide viable alternatives to the multilayer feedforward neural network for real-world applications.