What may happen or what you should do? Effects of knowledge representation regarding necessity of intervention on driver performance under level 2 automated driving

摘要

Drivers of SAE Level 2 systems are expected to perform the Object and Event Detection and Response (OEDR) task even when the driving automation system is working. An ideal Human Machine Interface should dynamically inform the driver of all of the system's functional limitations and driver's role. However, systems currently available in the market rely to some extent on such information given to the driver as knowledge at a dealer shop prior to driving or through mass-media. Since information acquirable as knowledge by users is limited, it is necessary that the information is given in a simple and comprehendible form. In this study, we compare user-oriented and system-oriented expressions for the information that explains that there are cases where a driver needs to intervene. This study conducted an experiment with a high-fidelity driving simulator where the ego vehicle was equipped with a Level 2 driving automation system. The results of the experiment suggest that expressions such as “you might need to intervene” (i.e., this is explained from the viewpoint of the driver) are better than expressions such as “the system may go wrong” (i.e., this is explained from the viewpoint of the driving automation system).