Knowledge based error recovery in industrial robots

This paper describes the main aims of a new research project concerned with the implementation of automatic error recovery facilities in industrial robotics. An approach is discussed in which an existing manufacturing work cell is to be enhanced by the addition of a task event model contained in an error recovery knowledge base. This paper outlines the main design issues involved in this work. Reasons why conventional fault tolerance techniques are inadequate are given and the industrial application is explained. A particular approach to sensory monitoring and error diagnosis is described. The proposed system has more similarities with sensory driven expert systems than with body modelling contingency planners.