Impediments to Environmental Awareness in Autonomous Driving Systems and its Effect on User Adoption

User adoption and technological advances are crucial for the widespread implementation of Autonomous Vehicles (AVs) in the existing transportation system. This study focuses on user adoption, with respect to the skepticism towards the environmental awareness of Autonomous Driving Systems (ADS). It leverages emerging simulation technology and examines the perceptual differences between ADS and human riders in high-level autonomous driving scenarios. Based on a virtual-reality-enabled human-in-the-loop experiment, it is found that human drivers (1) tend to perceive their roles as “drivers” rather than as “passengers” during autonomous driving, in contrast to people who do not know how to drive; (2) demonstrate a heightened sensitivity to traffic environments, including traffic signals and the movements of surrounding and cutting-in vehicles, compared to the ADS. Clearly there exists inconsistency between humans and ADS in environmental awareness, which leads to rider interventions and consequently affects user adoption of the AV technology. Furthermore, the gaze patterns of the riders are collected and analyzed during the experiment. A strategy using visual and audio cues is developed and implemented, so as to prevent unnecessary interventions. It is found that these cues are indeed helpful and decrease the number of interventions. The insights and methodologies are critical for improving the fundamental intelligence systems integral to autonomous driving.