Object manipulation under disturbance using fuzzy-neural network

Manipulating an object is one of the most important tasks of robots. A lot of studies have been done on object control algorithms and optimizing control force. Most of these studies, however, deal with objects whose properties are previously known and do not take into account disturbances. In this paper, an object control method under disturbance is proposed using fuzzy-neural network. The controller consists of main controller for an object trajectory control and sub controllers for manipulator force control to an object. Since manipulators have to apply commanded force to an unknown object, over-shooting might happen if the object is much harder than previously estimated. Therefore the environment estimator is set in the sub controllers in order to avoid big over-shooting which might give some damage to the object. Computer simulation was done under three types of disturbance to evaluate proposed control.