Counteracting the negative effects of high levels of train automation on driver vigilance

In a previous train simulator experiment (exp. 1), university student drivers, who were confronted with rare safely critical events, had a longer delay initiating emergency braking (inferring reduced vigilance) when the train had a high Level Of Automation (LOA) - an autopilot -, compared to when the train had a nil, low, or intermediate LOA (all manually driven trains). In this paper, the designs of two follow-up experiments are described. The aim of experiment 2 is to determine the effects of increasing task demands for train autopilot supervisors, beyond that of the same group in experiment 1, on mental workload. The aim of experiment 3 is to compare two strategies that attempt to counteract the negative effects of high LOAs on driver vigilance. We predict an improvement in the vigilance of drivers who, in addition to the autopilot supervision task, also have to respond to a many (60) vigilance device prompts over the 70 minute journey. Both a sensory and a cognitive vigilance device are to be tested for their comparative effectiveness. The sensory vigilance device (similar to that used in trains currently) requires drivers to press a single button in acknowledgement of an audible tone. The cognitive vigilance device (an experimental prototype) requires drivers to choose and press one of nine numbered buttons, in response to a maths quiz that relates to upcoming speed signs and signal aspects. The predicted results are discussed with reference to theories of mental work and vigilance task performance.