Abstract
Robots that explain something to people are actively researched for the use of museum guide, etc. However, the flexibility of interactions was limited because these robots were often implemented as non-human characters such as stuffed animals. Moreover, multi-user support and bidirectional communication with users were difficult with these robots. In this paper, we aimed to make a humanoid robot to conduct effective presentations by using its body gestures, hand gestures, and virtual laser pointers, as done by human presentators. Observing users through cameras, the robot should change its behaviors according to the states of the users, e.g. the robot should provide a more detailed explanation if people's understanding seems to be insufficient. This paper describes the development of our presentation robot system, and reports several evaluations conducted with small audience groups.
