Speech input applications for driving: using different levels of fidelity in simulator research

Transport simulators provide the potential to isolate specific variables to test for differences in controlled experiments. If simulators operate on a spectrum from low-end, paper based schematics, through to high-end full replica systems, there are fundamental issues associated with how ‘fit for purpose’ they might be. The physical and functional fidelity of simulators are key factors in their successful implementation as well as issues of face validity and ecological validity (the extent to which research findings generalise from one setting to another). This paper explores different levels of fidelity across a number of experiments that were conducted as part of the UK Government LINK Inland Surface Transport (IST) programme that investigated the potential impact of speech input systems in automotive applications. The findings illustrate that high-end simulation is not always required, especially in exploratory studies where fundamental principles are identified for further examination. However, there comes a point where more realistic driving tasks and contexts are required to develop higher levels of fidelity and enhance the ecological validity of results and the application of knowledge beyond the laboratory and into the real world.