Abstract: Automated driving (AD) is one of the key directions in the intelligent vehicles field. Before full automated driving, we are at the stage of human-machine cooperative driving: Drivers share the driving control with the automated vehicles. Trust in automated vehicles plays a pivotal role in traffic safety and the efficiency of human-machine collaboration. Therefore, it is vital for drivers to keep an appropriate trust level to avoid accidents. Here we proposed a new dynamic trust framework for AD, elaborating core factors affecting trust at different stages of trust development. The framework distinguishes four stages of trust (i.e., dispositional, initial, ongoing, and post-task trust), which are predominately affected by three distinct factors, including driver, AD system, and driving-related situation.
Dispositional trust is the initial stage of trust development and represents a driver’s overall trust propensity. It is predominately affected by the driver’s inherent trait, which refers to stable biological or social characteristics of the drivers (e.g., personality, age, cultural background, etc.). Initial trust reflects drivers’ trust status when they begin to use the AD system. In this stage, drivers have already learned the knowledge about the AD system or the driving situation, and the trust level is predominately influenced by driver’s trust propensity and prior knowledge. The prior knowledge largely comes from others’ descriptions of the AD system and drivers’ previous experience with similar systems. When a driver is adopting the AD, ongoing trust functions and impacts driver’s AD-reliance behavior dramatically. Because the level of ongoing trust directly affects whether and the extent a driver adopts the AD system, the ongoing trust is the crucial stage that trust calibration must pay attention to. During driving, the objective factors regarding the AD system (e.g., capability of AV system, the process of the AD system in fulfilling a task) and driving situation (e.g., weather, road conditions) directly influence the AD performance. However, the objective factors do not directly affect ongoing trust, but are transformed into subjective property through the driver's cognitive system. Subjective property hence is the direct factor affecting ongoing trust. The appropriate level of ongoing trust depends on driver's accurate comprehension of these objective factors. Meanwhile, the driver factor also affects the ongoing trust, by exerting a direct impact over ongoing trust and modulating the process of transforming objective factors into subjective property. Finally, post-task trust is a post-hoc evaluation of ongoing trust after completing the overall driving task. Factors affecting post-task trust hence largely overlap with the ones affecting ongoing trust.
According to the dynamic trust model, ongoing trust is the key to successful trust calibration. To achieve this aim, trust calibration could be achieved from the following three ways. First, the system monitors the drivers’ trust state and offers proper interventions in time when the drivers under- or over-trust the AD system. Second, training the drivers so that they have a more accurate mental model of the AD system. Third, since the human-machine interface (HMI) plays a key role in raising the transparency and understandability of system factor and situation factor, optimizing the HMI design is an important manner. In this way, driver’s transformation of objective factors into subjective property is improved, and a higher-level situation awareness is reached for the driver. Finally, basing on the dynamic trust framework as well as existing studies, we consider that the following six perspectives can be explored in future studies: (1) the influence of driver’s characteristics on trust, (2) the influence of HMI design on ongoing trust, (3) the sensitive measurement of ongoing trust, (4) the functional specificity of trust, (5) the mutual trust mechanism between the driver and AD system, and (6) improving the external validity of trust research.

Key words: trust, automated driving, dynamic trust framework, trust calibration, HMI design