Rephrasing the Problem of Robotic Social Navigation

We put forward the idea that to achieve robotic social navigation, it is necessary to consider different points of view (e.g. passenger and passer-by). We explore this idea in the context of controlling an intelligent powered wheelchairs in an environment with other pedestrians. We present results for a pilot study showing that the assessment of robot behaviour is dependent on these points of view. Failure to consider all potential actors could lead to situations of social awkwardness and the alienation of the primary user or others in the environment. Finally, we propose to examine predictability as a valid indicator of subjective experience and describe a theoretical model that would allow to identify features of unpredictable movement from different perspectives.