Development of Transparency Mechanisms to shape Functional Mental Models for improved Human-Robot Interactions

The applicability of robots significantly changed over the last years. Instead of static environments in the industry, robots are utilized in highly dynamic environments. Additionally, the interaction between humans and robots emerged as an important factor for the applicability of robots in personal environments. While considerable research was already invested to allow for more natural interactions with robots, errors occur regularly. Besides technical failures of the robot, human-induced mistakes greatly affect occurring problems. These mistakes can be ascribed to the unfunctional mental model naive users have about robots. In this context, an unfunctional mental model denotes false expectations of the functionality of a robot. This thesis tackles the problem of erroneous interactions by shaping functional mental models. The main hypothesis is that through more accurate assumptions about the functionality of robots, users will induce fewer errors in interactions. Towards this, approaches to convey the technical concepts of the robot are investigated. In a first step, I discuss the current literature on Human-Robot Interaction. A particular focus is on factors that influence users’ mental models of robots. Furthermore, approaches to increase robot transparency are presented. Based on this background, I investigate the effects of instruction videos and process visualizations on the mental model in an online user study. From the analysis of this study, I present the development of a transparent robot control architecture called FunctionAl user Mental model by Increased LegIbility ARchitecture (FAMILIAR). This control architecture aims at increased transparency and comprehensibility for naive users. I further investigate in a second study how dynamic visualizations and a visual programming approach can improve knowledge about the control architecture. The studies reveal that 1) the combination of an instruction video and a process visualization improves the mental model most successfully, 2) technical concepts of robots differ in terms of familiarity and perceivability, 3) functional mental models about robots improve Human-Robot Interaction. Additionally, observations regarding the negative impact anthropomorphization has on successful interactions is discussed.

Level thesis:degree_level

thesis.doctoral

Grantor dc:publisher

Universität Bielefeld

Year

2022

Author dc:creator

• Hindemith, Lukas

Repository record source_url

https://pub.uni-bielefeld.de/record/2961958

OAI identifier oai:identifier

oai:pub.uni-bielefeld.de:2961958