Concerns have been expressed regarding the validity of results obtained using non-drivers when conducting driving-related research in train simulators, yet it can be difficult to recruit experienced train drivers to take part in studies. In contrast, test participants, who lack train driving experience, are readily available. It has been suggested that, following a focussed, but relatively short, period of training/familiarisation, novice drivers may be sufficiently skilled at driving a train within a simulated environment for the purpose of taking part in a research study, even though they will clearly still lack the highly specialised skills, knowledge of rules and experience required to command a train on a conventional rail network in all manner of events. Early data (Dunn & Williamson, 2012; Large, Golightly & Taylor, 2014) appear to support this proposition, indicating that trained novices exhibit many similar patterns to experienced drivers. For example, a research study conducted by the authors, involving 13 novices and 7 experienced train drivers, investigated the driving performance and opinions of participants while utilising in-cab driver advisory systems. Results showed consistency between groups (experienced drivers versus novices) regarding driving performance (as determined by exiguous measures such as speed limit exceedances) and some subjective workload measures (obtained using NASA-TLX questionnaires), suggesting that the behaviour of non-drivers may be equally as valid as that of experienced drivers for this type of research.
However, performance measures, such as speed exceedances, are both coarse and occur infrequently, thus lacking statistical power. The concern is that although these metrics may indicate adequate performance, the manner in which that performance is effected (i.e. the control of the train) is fundamentally different between expert drivers and trained novices. To examine this, the driving performance data from the Large et al (2014) study was analysed using more sensitive measures, such as acceleration noise and the frequency and duration of control actions. This analysis exposed significant differences between groups, particularly pertaining to train handling strategies and techniques. Trained novice drivers (mainly comprising students and staff at the University of Nottingham) delivered erratic speed profiles, characterised by cruder control actions and high acceleration noise. In contrast, experienced drivers made more subtle refinements to power and braking force (higher frequency, shorter duration applications) achieving smoother speed control and lower acceleration noise.
Overall, the results of this further analysis suggest that while non-drivers may provide a valuable resource during rail driving research in a simulated environment, and general measures obtained from this group remain valid, care should be taken inferring behaviour or drawing specific conclusions relating to train handling strategies and techniques, such as speed control/driving performance, particularly when relying exclusively on a cohort of non-drivers. This study also highlights the utility of using metrics such as acceleration noise to provide a more sensitive assessment of train driving performance.
Train driving models and performance , Traffic management and driver advisory systems
