This paper reports on work related to the modelling of Human-Robot Communication on the basis of multimodal and multisensory human behaviour analysis. A primary focus in this framework of analysis is the definition of semantics of human actions, i.e. verbal and non-verbal signals, in a specific context with distinct Human-Robot interaction states. These states are captured and represented in terms of communicative behavioural patterns that influence, and in turn are adapted to the interaction flow with the goal to feed a multimodal human-robot communication system. This multimodal HRI model is defined upon, and ensures the usability of a multimodal sensory corpus acquired as a primary source of data retrieval, analysis and testing of mobility assistive robot prototypes.