Children's knowledge and expectations about robots

Abstract

This paper seeks to establish a precedent for future development and design of social robots by considering the knowledge and expectations about robots of a group of 296 children. Human-robot interaction experiments were conducted with a Tele-operated anthropomorphic robot, and surveys were taken before and after the experiments. Children were also asked to perform a drawing of a robot. An image analysis algorithm was developed to classify drawings into 4 types: Anthropomorphic Mechanic/Non Mechanic (AM/AnM) and Non-Anthropomorphic Mechanic/Non Mechanic (nAM/nAnM). Image analysis algorithm was used in combination with human classification using a 2oo3 (two out of three) voting scheme to find children's strongest stereotype about robots. Survey and image analysis results suggest that children in general have some general knowledge about robots, and some children even have a deep understanding and expectations for future robots. Moreover, children's strongest stereotype is directed towards mechanical anthropomorphic systems.