Framework for Understanding the Driver's Trust in Automation and Its Implications on Driver's Decision and Behavior
Author(s)Cho, HongSeok, R. John Hansman
The aim of this paper is to understand how designs of the automated driving system influence the driver's automation use decisions, and to provide design recommendations for SAE level 2 or 3 automated driving systems that would promote appropriate decisions of the driver. A risk-based framework, "decision matrix" is developed to represent the driver's decisions in terms of two variables: 1) perceived reliability of the automation, and 2) perceived consequence of the automation's potential unreliable behavior. Each block of the matrix represents a level of the perceived risk and an accompanying automation use decision of the driver; 1) use, 2) use with monitoring, and 2) do not use. Having the decision matrix at the core, an overall driver-automation system architecture is developed in order to describe how the driver makes a cognitive assessment of the observable states to evaluate the two decision matrix variables. The architecture also includes a learning process by which the driver develops and evolves his/her mental model based on the experience. The architecture is used to identify potentially inappropriate decisions of the driver as a result of the inaccurate evaluations of either the automation reliability or the consequences, or both. The framework suggests various methods (e.g. minimum performance of the automation, limiting automation use, information display and interface design, and training) which manufacturers can design to support the driver in learning of the automation's limitations and thus making appropriate evaluations of the decision variables. As a next step, potential designs of these methods to promote appropriate automation use decisions will be evaluated to determine the effectiveness of the methods.