This work aims to advance the state of the art in exploring the role of task, social context and their interdependenciesintheautomaticpredictionofaffectiveandsocial dimensionsinhuman–robotinteraction.Weexploredseveral SVMs-based models with different features extracted from a set of context logs collected in a human–robot interaction experiment where children play a chess game with a social robot. The features include information about the game and the social context at the interaction level (overall features) and at the game turn level (turn-based features). While overallfeaturescapturegameandsocialcontextattheinteraction level,turn-basedfeaturesattempttoencodethedependencies of game and social context at each turn of the game. Results showed that game and social context-based features can be successfully used to predict dimensions of quality of interactionwiththerobot.Inparticular,overallfeaturesprovedto performequallyorbetterthanturn-basedfeatures,andgame context-based features more effective than social contextbased features. Our results show that the interplay between game and social context-based features, combined with featuresencodingtheirdependencies,leadtohigherrecognition performances for a subset of dimensions.

EB00000003683K

Available

• \Detecting perceived quality of interaction with a robot using contextual features