Allocation of Function to Humans and Automation and the Transfer of Control

Following a short description of function allocation (FA), and its use in other domains, this chapter provides a summary of the likely human factors challenges faced by system engineers and designers currently developing and assigning different functionalities for automated vehicles. We argue that, historically, allocation of particular functions to a machine has been motivated by the wish to relieve the user of monotonous, repetitive, or unsafe tasks, or for providing system capabilities that are faster, stronger, or more capable than humans. The rationale for allocating more tasks to the vehicle, and increasing the likelihood of higher level of automated driving, is primarily based on the desire to reduce the number of human-based errors and limitations, which are known to lead to crashes, and reduce road safety. Well-designed automated vehicles also promise to reduce transport-related emissions and congestion, and pledge increased productivity, by relieving the humans from the monotonous and repetitive driving task, affording them the opportunity to perform other tasks. However, we provide evidence from studies which suggest that, due to the complex and dynamic nature of driving, and the continually fluid responsibility for tasks between the system and the user, there remain some fundamental challenges for system designers in this domain. For automated vehicles to deliver on their promises, and reduce transport-related crashes, it is imperative for engineers and designers to be aware of the unintended consequences of human interaction with, and expectations of, their (almost perfect) system. In addition to increasing the likelihood of new errors; inappropriate, untimely, or prolonged allocation of function to the system has been shown to lead to user confusion, distraction, fatigue, loss of skill, and complacency, which have ultimately led to problems with the transfer of control, when the technology reaches its limitations. An important, and yet ill-understood area of research in this context involves better ways of communicating system capability to the user, ensuring they have the correct mental model, which may in itself require regular updates. Keeping the driver vigilant during prolonged periods of system use and understanding what sustains their ability to resume control from the machine are other areas which require further knowledge in this context, as are considerations of how to manage failed transfer of control. We argue that, once more knowledge is acquired in these areas, manufacturers and authorities can work towards providing better training protocols for users, and developing better standards for Human–Machine Interfaces, to ensure highly automated vehicles deliver on their aspired promises.