Özet:

Humanoid robots are extensively becoming an essential part of the social life.  It is crucial for humanoid robots to understand the emotions of the people for efficient human-robot interaction. Even though a great number of facial emotion analysis algorithms have been developed and a number of them have been implemented to humanoid robots, there are still gaps in improving accuracy, computational burden and speed of these algorithms.  This paper proposes a 4-stage emotion analysis algorithm and then presents its application to NAO humanoid robot. Initially, the robot detects the face using Viola-Jones algorithm. Later, important facial distance measurements are taken with geometric based facial distance measurement technique. Then, facial action coding system technique is used to detect movements of the measured facial points. Finally, measured facial movements are evaluated to understand instant emotional properties of the person. Although this algorithm can be implemented to all humanoid robots, in this research, it has been specifically applied to NAO humanoid robot. The reliability of the emotion analysis is verified by analyzing each terminal decision made based on the facial distance measurements. In addition, the accuracy, computational burden and speed of the algorithm are assessed to show the effectiveness of the algorithm. 

Anahtar Kelimeler: