Fully autonomous vehicles can be classified as robots. In this paper we propose to approach the development of autonomous vehicle user interfaces from a Human-Robot Interaction perspective, based on two principles. First, different robots require different user interfaces depending on their level of automation. Second, as the level of robot automation increases so should the automation of the interface itself; creating a spectrum ranging from a conventional “master-slave” level interaction to a fully intelligent “equal-footing” level interaction. Two research questions arise: where along the spectrum described above should autonomous vehicle user interfaces be, and what technological advance would have the great-est impact in enabling those interfaces. This paper presents the theoretical foundation of our research at the intersection of three previously unconnected fields: autonomous vehicles, human-robot interaction and affective computing. We then outline an experimental framework for developing a prototype interface based on our findings.