Virtual Human Characters for Autonomous Vehicle-to-Pedestrian Communication

Pedestrians base their street-crossing decisions on both vehicle-centric cues, like speed and acceleration, and driver-centric cues, like gaze direction and facial expression. In the future, however, drivers of autonomous vehicles will be preoccupied with non-driving related activities and thus unavailable to provide pedestrians with relevant communicative cues. External human-machine interfaces (eHMIs) hold promise for filling the expected communication gap by providing information about the current state and future behaviour of an autonomous vehicle, to primarily ensure pedestrian safety and improve traffic flow, but also promote public acceptance of autonomous vehicle technology. The aim of this thesis is the development of an intuitive, culture-transcending eHMI, that can support multiple pedestrians in parallel make appropriate street-crossing decisions by communicating pedestrian acknowledgement and vehicle intention. In the proposed anthropomorphic eHMI concept, a virtual human character (VHC) is displayed on the windshield to communicate pedestrian acknowledgement and vehicle intention via gaze direction and facial expression, respectively. The performance of different implementations of the proposed concept is evaluated in the context of three monitor-based, laboratory experiments where participants performed a crossing intention task. Four papers are appended to the thesis. Paper I provides an overview of controlled studies that employed naive participants to evaluate eHMI concepts. Paper II evaluates the effectiveness of the proposed concept in supporting a single pedestrian or two co-located pedestrians make appropriate street-crossing decisions. Paper III evaluates the efficiency of emotional facial expressions in communicating non-yielding intention. Paper IV evaluates the efficiency of emotional and conversational facial expressions in communicating yielding and non-yielding intention. An implementation of the proposed anthropomorphic eHMI concept where a male VHC communicates non-yielding intention via an angry expression, cruising intention via cheek puff, and yielding intention via nod, is shown to be both highly effective in ensuring the safety of a single pedestrian or even two co-located pedestrians without compromising traffic flow in either case, and the most efficient. Importantly, this level of effectiveness is reached in the absence of any explanation of the rationale behind the eHMI concept or training to interact with it successfully