Development of Human-Robot Communication Technologies for Future Interaction Experiments

• Ana Christina Campos UFMG
• Bruno Adorno UFMG

Resumo

This paper describes our development of humanrobot communication technologies, which will be used in future experiments to investigate the effects of different types of communication in human-robot interactions. Using the human kinematic chain and facial points, obtained with methods from the literature, we propose geometric analyses for recognition and interpretation of human pointing gestures and gaze direction. We developed a virtual agent with voice, facial expressions, and gaze direction to interact with humans. The pointing gestures system presented mean success rate of 77.8% (s.d. 16.4%) in experiments and the implementation of the facial points detection algorithm resulted in mean errors of 0.52 cm (s.d. 0.29 cm) in the horizontal iris centers distance and 0.67 cm (s.d. 0.31 cm) in the vertical distance between eyes and mouth, both important attributes to estimate human gaze direction. Those results are satisfactory for the intended application.

Palavras-chave: Robots, Iris, Mouth, Hip, Elbow, Cameras, Robot kinematics