Abstract

In a society that keeps getting closer in touch with social robots it is very important to include potential users throughout the design of the systems. This is an important rationale to build robots that provide services and assistance in a socially acceptable way and influence societies in a positive way. In the process, methods are needed to rate the robot interaction performance. We present a multimodal corpus of naïve users interacting with an autonomously operating system. It comprises data that, to our conviction, reveal a lot about human-robot interaction (HRI) in general and social acceptance, in particular. In both, the evaluation and the design process we took into account Clarkson and Arkin's heuristics for HRI (developed by adapting Nielsen's and Scholtz' heuristics to robotics) 1. We discuss exemplary results to show the use of heuristics in the design of socially acceptable robots.