Abstract: This thesis is focused on context aware interaction of human with a smart environment. It analyses and exploits the fusion between real and virtual world, in order to propose an innovative interaction paradigm and model. The proposed model will be the bridge to fill the gap between physical and virtual world and take advantage of mixed environments in order to facilitate and enhance human capabilities. Research axes are related to interaction (egocentric, mixed reality, multimodal, and hybrid) and to context awareness (multimedia engineering, sensors and actuator). In particular, ARAMIS takes in account gesture human-machine interactions with an egocentric perspective. This perspective is based on a situative model of what a specific human agent can see and not see, reach and not reach at any given moment in time. The particularity of this model is that physical and virtual objects are treated as being located in the same space. In order to obtain this concept ARAMIS will exploit several technologies from computer vision to sensor network, from wearable technologies to smart environment.