Sheep and Wolves Testbed for Interaction and Collaboration between Humans and Robots

Abstract

This paper presents the first prototype of Sheep and Wolves, a system for testing interaction and collaboration paradigms between humans and robots. The paper contributions are twofold: a mixed reality interface for human-robot interaction, and a practical experimental tool for assessing how different robotic behavioral patterns affect interaction and collaboration with users. Sheep and Wolves places humans, robots and virtual entities in a game environment where they have to collaborate and compete. The system is designed around the classic Sheep and Wolves board game, played on a large physical checkerboard. In the prototype presented here the user is playing a single wolf in a pack of four autonomous robotic wolves trying to hunt a single virtual sheep. The human interacts with the rest of the wolf pack using a mixed reality video stream, a graphical interface and a text chat tool that enables discussion and planning of future moves within the pack. In preliminary testing Sheep and wolves was sensitive to differences in the robots behavioral patterns and suggested that robotic assertiveness (or robotic chutzpah) might enhance the quality and trustfulness of the interaction.