This paper addresses the problem of creating general navigation systems that would apply to different types of autonomous vehicles, including indoor and outdoor robots, with different physical dimensions and sensor characteristics. The authors outline a neural network approach to navigate when there are obstacles and targets. The paper describes successful experiments involving the outdoor Robucar and the indoor Robuter.
The exposition requires the reader to have prior knowledge of the critical terminology and concepts of neural networks. References are given to a book and articles, along with three other papers by the authors, but it would have been better if this paper were more self-contained. In addition, the writing is not standard, idiomatic English. The underlying work is impressive because of its systematic approach to a variety of settings and it deserves the best possible exposition.