Intelligent assembly in flexible automation
This work investigates the automation of assembly cells and the need to incorporate sensor-guided decision techniques. The experience of industry in this area is examined by observing a real cell on the shop floor. From the collected data conclusions point to an alternative error interpretation which describes the successful completion rather than an enumeration of errors. A methodology for the description of the process in robotic assembly is developed. The constituent phases in handling components are identified as Feeding, Transport and Mating. Each phase has well defined characteristic properties which can be determined using appropriate sensing mechanisms. The mating phase is given special attention by proposing the method of information Spaces as a suitable frame work for sensor fusion and context directed interpretation. Thus the successful progress is described regarding any deviations as errors. They in turn can be interpreted in the context in which they were encountered and recovery is accomplished in the demonstration cell by operator taught routines. Where error repetition occurs, a simple look-up technique suffices to remove the need for another operator intervention. The required data structures and the implementation of the experimental cell are discussed. It is concluded from the results that the principle of knowledge-based assembly control exhibits an intelligent behaviour which contributes to an increase in the cell productivity. This method addresses only a part of the overall problem of assembly automation, but it has a central place in the system and could be extended to the complete system.