Facial Recognition and Pathfinding on the Humanoid Robot Pepper as a Starting Point for Social Interaction

Abstract

Interaction between humanoid robots and humans is a complex process. Speech, gestures, and recognition of communication partners are important aspects in a well-defined interaction. To seem more natural, a humanoid robot should not be stationary. It should be able to be part of a crowd and wander around a specific area. Therefore, pathfinding is important to give a humanoid robot the ability to connect with people in more than one place. In addition, the recognition of communication partners is the backbone of social interaction. This chapter demonstrates how OpenCV, a well-known computer vision library, supports the robot Pepper in the recognition of communication partners and in addition, how this is the starting point to different types of small talk as the basis for a prototypical interaction process of humanoid robots and humans. Additionally, the navigation functions that allow the robot to move autonomously and enable a better human–robot interaction will be discussed.