Automated driving technology is developing at a steady pace, and is expected to permeate the society in the near future. It brings with itself the several promises such as increased traffic flow, better safety, and mobility for people with special accessibility needs. However, even if automated driving technology becomes mature for large scale deployment, it will be the human factors that dictate the success of the integration of automated vehicles (AV) in traffic and society. Public perception of automated driving technology varies considerably, ranging from enthusiasm and curiosity, to distrust and antipathy. In the context of this hesitation, sharing road spaces with automated vehicles – particularly in the nascent stages of this technology – may not happen organically. The design of a good AV-pedestrian interaction paradigm should consider the user perspective, and research is needed to understand the perceptions and needs of pedestrians in order to contribute to a smooth transition to and a continued success of automated driving. It is commonly believed that there is explicit interaction between drivers of manually driven vehicles and pedestrians. Taking the driver out of the loop – as in Automated Driving – takes away this explicit interaction. This raises the question of how this communication gap can be addressed. The commonly proposed solution is to extend AVs with an external Human-Machine Interface (eHMI) to enable explicit AV-pedestrian interaction.

This research looks into the interactions between automated vehicles and pedestrians and aims to inform the design of an eHMI. The research aims to identify the aspects of interaction that drive road-crossing behaviours, and to uncover attributes of external communication through an eHMI that effectively facilitates this interaction. The work starts by identifying the theoretical requirements of a successful eHMI through a conceptual analysis and organizes the currently diverse space of eHMI concepts by describing a taxonomy and classifying said concepts. Subsequently, the work addressed six research questions in the context of pedestrians’ interaction with vehicles in a road-crossing situation: (1) How critical is explicit communication in traffic interactions at present? (2) Where do pedestrians look to get the necessary information and make road-crossing decisions in front of an approaching vehicle? (3) How do pedestrians’ interactions with AVs di er from that with ordinary, non-automated vehicles? Much of related research in the field assumes that an eHMI is a good solution to bridge the communication gap between AVs and pedestrians. To this end, we ask: (4) What is the contribution of an eHMI in facilitating a smooth

interaction, and what role does the vehicle’s driving behaviour play? (5) What are the user preferences for colour and animation patterns in an eHMI? And particularly given that pedestrians seek different information based on the distance of an approaching vehicle, we ask: (6) What are the merits of phase-by-phase, distance-dependent communication of a yielding AV’s intent? These questions were answered using a mix of methods commonly used in design research. The research culminates in evaluations and recommendations for properties of an external Human-Machine Interface (eHMI) for Automated vehicles.