Synthesis of Relevant Communication Features in Vehicles with Automated Driving Systems

Liam Kettle

Advisor: Yi-Ching Lee, PhD, Department of Psychology

Committee Members: Elizabeth Phillips, Patrick McKnight

Horizon Hall, #1008 (ALC) and Online
June 28, 2024, 01:00 PM to 03:00 PM

Abstract:

Vehicles equipped with automated driving systems (ADS) partially relieve human drivers of vehicle control until the ADS has full control of driving actions and decision making. However, concerns arise regarding the transparency of the ADS’ actions, decision making, and situation awareness (SA), contributing to drivers’ inappropriate trust and reduced SA. One way to mitigate transparency concerns is to communicate relevant actions, decisions, and road elements to the driver.  To synthesize relevant communication features, three studies were conducted. The first study systematically reviewed in-vehicle augmented reality (AR) interfaces and identified heterogenous communication systems and the need for understanding relevant, scenario-specific communication features. The second study explored drivers’ information needs across various driving scenarios. Participants watched a series of driving scenarios and provided feedback on desired and redundant information. This resulted in a framework of key vehicle-human communication features that would enhance drivers’ anticipated trust and SA. This framework identified scenario-stable and scenario-specific features supporting the importance of only communicating relevant information. The third study validated the findings from study 2. Participants watched a series of driving scenarios with different communication interfaces corresponding to four levels of information relevancy (no cues, relevant cues, full cues, re-test of no cues). The findings indicated that self-reported trust (general and situational) and usefulness (but not satisfaction) of the ADS was greater than presenting no information but did not differ across relevant and full cue conditions. This indicates that communicating greater transparency (full cues) does not offer additional perceptual benefits and that presenting relevant cues was more efficient at enhancing individuals’ perceptions. Overall, the three studies provide a step-by-step approach in reviewing, exploring, and validating a driver-centric vehicle-human communication framework of features considered relevant by drivers. The findings build a foundation for standardizing which vehicle and environment features are relevant and when they are relevant across driving scenarios. The vehicle-human framework is expected to foster more consistent ADS communication research that ultimately improves driver safety through enhanced system transparency.