Investigating Naturalistic Ways for Drivers to Intervene in the Fully Autonomous Vehicle System while Performing Non-Driving Related Tasks

Abstract

With increasing acceptance for autonomous vehicles (AV), drivers turn their primary attention to what would otherwise be secondary tasks (e.g., eating, using a smartphone, or working on a laptop). These activities make user’s behavior and concept towards their vehicle will dramatically change. Thus, a new paradigm of interaction will appear with the autonomous vehicle which requires novel, easy, and reliable interaction inputs to facilitate communication between the driver and the AV. However, in Fully Autonomous Vehicles FAVs many studies have investigated the proper input interactions that help the passengers to communicate and order a mission from the vehicle. These suggested input interactions however were not based on considering the differ of the non-driving related activities (physical or cognitive) or even considering the type of the intervention scenario, and thus no consensus on best input interactions has emerged. To address this gap, in study1 (32 participants), we compared three input interactions, a new suggested input in AV (gaze-head input) with the commonly used inputs (touch and voice inputs) to provide a holistic view about the effect of involving in NDRT (using a smartphone) and the differ of the driving scenarios on the preferred input interaction. We found that the NDRT effected on the preferred input interactions and the difference of the driving scenarios were affected on the reaction time of tested inputs. Accordingly, in this study, we investigate the factors that influence on driver’s natural interaction with the FAV system. We presented an online video-based survey to 360 participants showing four conditions of NDRTs based on different physical and cognitive engagement levels, tested with six of the most common intervention scenarios (24 cases) with seven input interactions for each case: touch, voice, hand-gesture, and their combination. Results conclude that NDRTs have the highest influence on the driver’s input interaction than the intervention scenario categories. In contrast, the variation of the physical load has more influence on the selection than the variation of the cognitive load.