ABSTRACT
As driver assistance and support systems become more complex and more varied, the potential for unanticipated behavioural response increases. One solution is to conduct extensive real-world trials, commonly known as field operational tests (FOTs), to evaluate such systems. But these trials are very costly and it is not practical to conduct them for every system, function, and variant of a function. Many FOTs are carried out on production versions of systems, meaning that feedback on problems will come only after products are already in real-world use. FOTs can be conducted on prototype systems, such as those conducted on intelligent cruise control (Fancher et al., 1998), on automotive collision avoidance systems (University of Michigan Transportation Research Institute and General Motors, 2005), on road departure crash warning systems (LeBlanc et al., 2006), and on integrated vehiclebased safety systems (Sayer et al., 2010, 2011). But, depending on the design of the study and the types of drivers recruited, it may not be possible to cover a range of driver types and personalities. In addition, prototype systems may be less reliable, which can, for example, result in spurious warnings leading to driver annoyance (e.g., Sayer et al., 2011).
