Vehicle automated driving systems, capable of performing all dynamic driving tasks are profoundly changing the traffic environment. With increased automation the vehicles are gradually becoming artificial agents that act and interact in conjunction with human agents, creating a pristine social context. This paper addresses the interaction between ‘pedestrian-vehicle’ and ‘driver-vehicle’. The idea is to make interaction as humanlike as possible, to increase safety and a positive user experience. We suggest that concepts describing social attunement in human-human interaction also can be applied to human-vehicle interaction. Social attunement implies a scenario where human and vehicle share intentions, infer goals of the interaction partner, are mutually predictable, and understand performance limitations. Furthermore, we propose the use of an in-vehicle avatar interface to assist in this interaction. A reason for using an avatar is that it can be hypothesised that the user can interact more naturally with an anthropomorphic artificial agent.