Modelling and analysis of CAD expert behaviour in using manual input devices

In computer-aided design (CAD) systems the tablet-plus-keyboard has been th e most common configuration of input devices. Their use in CAD work require s several additional subsidiary tasks of device manipulation, decision-maki ng, perceptual matching and error correction. These tasks, imposed by the s ystems design, are mostly additional to the primary task of drawing. This o bservational study documented the nature of non-optimal behaviour in using manual input devices to perform CAD tasks by analysing behaviour protocol o f designers at work recorded on video. A task analysis illustrated the vari ety of knowledge resources in CAD performance. The use of manual input devi ces resulted in substantial movement of the eyes between different parts of the system, thus reducing the time spent looking at the graphics screen fo r design activity; and substantial movement of the hand between input devic es, thus incurring time. The imposition of subsidiary tasks is considered n on-optimal since more resources were deployed to manipulate the input devic es and correct errors than to perform mental operations required by the dra wing task. A blackboard model of system behaviour illustrates individual di fferences in knowledge recruitment due to different tasks and systems used. The model, based on behaviour protocols, provides a novel approach to task analysis of CAD expert performance. Relevance to industry The use of CAD systems in manufacturing is commonplace. It is important tha t these systems are designed to meet CAD designers' needs and to support th eir task performance. The study has implications for the use of speech inpu t devices in CAD which complement manual input and enhance the overall prod uctivity of CAD output. (C) 2001 Elsevier Science B.V. All rights reserved.