The ability to localise a smart device is very useful to visually or cognitively impaired people. Localisation-capable technologies are becoming more readily available as off-the-shelf components. In this paper, we highlight the need for such a service in the field of health and autonomy, especially for disabled people. We introduce a model for Semantic Position Description (SPD) (e.g. "The pill organiser in on the kitchen table") as well as various algorithms that transform raw distance estimations to SPD related to proximity, alignment and room identification. Two of these algorithms are evaluated using the LocURa4IoT testbed. The results are compared to the output of a pre-experiment involving ten human participants in the Maison Intelligente de Blagnac. The two studies indicate that both approaches converge up to 90% of the time. .