Classification of human touching behaviors using a whole body haptic interface for a pet-robot

摘要

We propose a classification method of human touching behaviors using a whole body haptic interface for a pet-robot. In our previous work, we have developped a classification method for five classes of human touching behaviors using a planer pressure sensor, which is difficult to be applied for a rounded shapes of the real robot. In this paper, we propose a new classification method using a new pressure sensor which are distributed into many particle units. Experimental results show that 76.8% classification ratio on the average for the five behaviors for the three body parts, front, back, and side. Classification results also show that the way of touching tends to differ between body parts even for the same individual.