A methodology for eliciting expert knowledge in virtual engineering environments

Summary form only given. This research was initiated and applied in an area of engineering where tacit knowledge predominates; that of manual assembly planning. In this domain the evaluation and formalization of tacit knowledge is very difficult. Assembly planning is generally undertaken by planning 'experts' who produce plans after studying design data in whatever form. Research at Heriot-Watt University has shown that immersive virtual reality (VR) offers significant potential for the interactive planning of mechanical assemblies. Using this system for gathering assembly intent, a knowledge elicitation methodology was developed in which expert assembly planners' activities were logged while building a virtual assembly. This involved four main stages in its application: definition of the salient component attributes (repertory grid analysis); evaluation of attributes for given components (manual evaluation); recording expert assembly usage data (nonintrusive VR data logging); and generating rules relating the components to specific assembly methods (induction). This paper demonstrates how this novel nonintrusive knowledge acquisition methodology can be applied using immersive virtual environments; implying that, as globalizes technology becomes available, this may be a route to the formalization of company know-how.