Fuzzy-neural approach in the development of cognitive robotic systems

Abstract

Several parallel advances have been made in the distinct disciplines: fuzzy logic and neural networks. As the names imply, the theory of fuzzy logic provides a mathematical framework for the emulation of certain perceptual and linguistic attributes associated with human cognition, whereas the science of neural networks provides new computing morphologies with learning and adaptive capabilities. A marriage between these two distinct disciplines has the potential of producing robotic machines with some sort of cognitive abilities. These cognitive systems will, hopefully, recapitulate certain aspects of human cognition such as learning, logic, thinking, perception, decision making in uncertain and unstructured environment, memory, etc. An integration of these two fields has a potential of producing robust sensors, and robust control mechanisms. We briefly examine these two fields: fuzzy logic and neural networks, and explore the possibilities of their integration in the development of cognitive robotic systems. Special emphasis is given to the vision and control aspects of robotics systems.