Emotional Postures for the Humanoid-Robot Nao

Abstract

This paper presents the development of emotional postures for the humanoid robot Nao. The approach is based on adaptation of the postures that are developed for a virtual human body model to the case of the physical robot Nao. In the paper the association between the joints of the human body model and the joints of the Nao robot are described and the transformation of postures is explained. The non-correspondence between the joints of the actual physical robot and the joints of the human body model was a major challenge in this work. Moreover, the implementation of the postures into the robot was constrained by the physical structure and the artificial mass distribution. Postures for the three emotions of anger, sadness, and happiness are studied. Thirty two postures are generated for each emotion. Among them the best five postures for each emotion are selected based on the votes of twenty five external observers. The distribution of the votes indicates that many of the implemented postures do not convey the intended emotions. The emotional content of the selected best five postures are tested by the votes of forty observers. The intended emotions received the highest recognition rate for each group of these selected postures. This study can be considered to be the last step of a general process for developing emotional postures for robots. This process starts with qualitative descriptions of human postures, continues with encoding those descriptions in quantitative terms, and ends with adaptation of the quantitative values to a specific robot. The present study demonstrates the last step of this process.