PUB - Publikationen an der Universität Bielefeld

Studies in human-robot interaction (HRI) typically involve computational artifacts, i.e. the robotic system, as the subject of investigation. Thus, the reproducibility of any result in HRI studies directly relates to the reproducibility of this computational artifact in the first place. This has certain consequences for appropriate workflows that will be discussed in this chapter. We argue for a higher awareness, improved standards, and further automation of tool chains used to conduct robotic experiments. We identify this as a research topic in its own right, especially in cases where robotic systems are used in interdisciplinary research. This inherently includes that technically complex robotic experiments should also be reproducible by scientists with a non-technical background. We analyze and discuss a dedicated study by the CITEC Central Lab Facilities and an international team demonstrating that it is possible to replicate a relatively complex HRI experiment in two different laboratories across the globe by a research assistant with no experience in robotics at all.