Mining consumer complaints to identify unsuccessful interactions with advanced driver assistance systems

Advanced driver assistance systems, which warn drivers of danger of an impending collision or temporarily control the vehicle's speed and/or direction under limited circumstances to assist the driver, have the potential to prevent large numbers of motor vehicle crashes, injuries, and deaths. Maximizing their potential safety benefits requires drivers to use them appropriately. However, drivers might misuse these systems if they overestimate the systems' capabilities. Driver might also disuse the systems if the systems fail to meet their expectations, diminishing the systems' benefits. This study seeks to identify unsuccessful driver interactions with advanced driver assistance systems by mining the text of a database of consumer complaints about vehicle safety issues. If successful, future work will attempt classify complaints reflecting faulty mental models versus possible system errors. This information can be used by industry to improve consumer education and driver-vehicle interface design and by researchers to guide future research needs.