Automated vehicles are gradually entering the market and the technology promises to increase road safety and comfort, amongst other advantages. An important construct guiding humans' interaction with safety-critical systems is trust, which is especially relevant as most drivers are consumers rather than domain experts, such as pilots in aviation. The successful introduction of automated vehicles on the market requires to raise the trust of technology skeptics, but at the same time prevent overtrust. Overtrust is already suspected of having contributed to a couple of - even fatal - accidents with existing driving automation systems. Consequently, there is a need to investigate the topic of trust in the context of automated vehicles and design systems which maintain safety by preventing both distrust and overtrust, a process also called "trust calibration". As the possibility to engage in non-driving related tasks is an important consumer desire, this work proposes to consider drivers' multitasking demands already in the vehicle design process to prevent emerging trust issues. Therefore, a framework integrating theoretical considerations from the domains of trust, human-machine cooperation, and multitasking is proposed. By aligning overall goals between the operator and the system whilst supporting drivers in tasks at the strategical, tactical, and operation level of control, a more trustworthy cooperation should be achieved. A series of studies was conducted to identify important dimensions of trust in driving automation as well as scenarios leading to distrust and overtrust. Those scenarios were then used to demonstrate how the structured approach provided by the framework allows for designing in-vehicle interfaces. Three interaction concepts aiming to support drivers in the different levels of automation were designed and evaluated in driving simulator studies. Results highlight the potential of multimodal as well as attentive user interfaces (interruption management) to deal with overtrust, and augmented reality visualizations to raise acceptance of drivers distrusting the automation. All approaches confirmed to improve the subjective trust of the operator and demonstrate the structured approach provided by the framework can assist to design more trustworthy in-vehicle interfaces, which is important for a successful and safe implementation of driving automation systems.
