Robot Behaviors for Transparent Interaction

Interacting physically and sharing the environment lead humans and robots to work with each other at close distance, and in such circumstances autonomous robots are expected to have safe and social behaviors. Thereby, developing a socially acceptable behavior for autonomous robots is a foreseeable problem for the Human-Robot Interaction field. We propose two methods for integrating transparent behaviors into social robots: a by-design and a by-learning approach. Furthermore, with this work we address the problem of transparency from the navigation point of view. Our first step has been to explore these two approaches by conducting a preliminary within-subjects study (33 participants). Our results showed that deictic gestures as navigational by-design cues for humanoid robots result in fewer navigation conflicts than the use of a simulated gaze. Additionally, the perceived anthropomorphism was increased when the robot used the deictic gesture as a cue. These findings highlight the need for a universal design approach to effectual non-verbal by-design behaviors to increase transparency in future humanoid robotic applications and also underscore the importance of investigating by-learning cues as the complexity of robots' internal state grows.