New hybrid control architecture for intelligent mobile robot navigation in a manufacturing environment

Abstract

This paper presents a new hybrid control architecture for Intelligent Mobile Robot navigation based on implementation of Artificial Neural Networks for behavior generation. The architecture is founded on the use of Artificial Neural Networks for assemblage of fast reacting behaviors, obstacle detection and module for action selection based on environment classification. In contrast to standard formulation of robot behaviors, in proposed architecture there will be no explicit modeling of robot behaviors. Instead, the use of empirical data gathered in experimental process and Artificial Neural Networks should insure proper generation of particular behavior. In this way, the overall architectural response should be flexible and robust to failures, and consequently provide reliableness in exploitation. These issues are important especially if one takes under consideration that this particular architecture is being developed for mobile robot operating in manufacturing environment as a component of Intelligent Manufacturing System.