Experiments were carried out within the University premises. Participants will be hosted in groups, but they will access to the assessment room one by one. To avoid the learning effect bias, the robotic and standard sessions will be spaced by at least two weeks. We will alternate which modality (robotic/standard) is tested first. A session is about 20 mins including welcome and debriefing. Details on the experiments are in the published articles. 
The audio and video data from interaction were used for refining the speech and object recognition via machine learning techniques.