Neuro-Fuzzy System for Robotics Applications

Abstract

Over the last two decades several advances have been made in the areas of fuzzy logic and artificial neural networks. It is interesting to note that fuzzy logic and neural networks complement each other, and their fusion provides the benefits of both the technologies. Neural networks can deal with imprecise data and ill-defined activities; thus, they offer low-level computational features. On the other hand, fuzzy logic provides higher-level cognitive features as it can deal with issues such as approximate reasoning and natural language processing. The merging of these two fields results in an emerging paradigm called ‘neuro-fuzzy systems’. These are believed to have considerable potential in the areas of robotics, expert systems, medical diagnosis, control systems, pattern recognition and system modeling. The intent of this paper is to describe a Neuro-Fuzzy System (NFS) for on-line computation of inverse kinematic transformations of a robot manipulator. The NFS is comprised of a conventional fuzzy logic controller in the feedback configuration and a recurrent neural network in the inverse mode (feedforward) configuration. It is envisaged that the integration of fuzzy logic-and neural networks—based systems will encompass the merits of both the techniques, and thus provide an intelligent and a robust mechanism. The NFS uses minimum fuzzy rules for a given task, thus, overcoming the problem of designing an elaborate Knowledge Base for the Fuzzy Logic Controller. The proposed scheme is implemented for computing inverse kinematic transformations of a two-linked robot. It should be noted, however, that the NFS is not to be considered as a ‘Fuzzy Neural Network’; but it is a synthesis of neural networks and fuzzy logic based control techniques.