COMPUTERS & OPERATIONS RESEARCH, cilt.18, sa.4, ss.343-353, 1991 (SCI-Expanded)
Artificial neural networks' ability to learn, categorize, generalize, and self organize make them potentially very useful for a variety of application areas. This paper discusses some experiences in attempting to teach a neural network how to perform the pure pursuit maneuver. Several problems were encountered in defining a proper training set. Three and four layer backpropagation networks were utilized to capture the input to output mappings.