Designing for interactions that enable end user impacts on robot behavior

Abstract

This dissertation is focused on how to design for interactions with robots in ways that enable end users to impact robot behavior. I consider the combination of interaction design and system architecture choices, and how these affect how end users interpret the technology and are able to act on and with the technology.Why consider interaction design of robotic systems? Different instances of robots can be programmed to do different things, even if they have the same embodiment. The embodiment of a robot may raise certain expectations regarding what the robot is capable of. For example, humanoid robots may raise expectations of humanlike interaction capabilities that the robot may not be able to meet. Often, models are borrowed from human-human communication and applied to Human-Robot Interaction by other researchers. These are symmetrical models that depict the human and the robot agent to have similar capabilities. I argue that we should instead see human-robot interaction as asymmetric, and that what the robot is (capable of) doing should be communicated through the robot’s physical and behavioral design. This can be done, for instance, by making the robotic system transparent, meaning that the system makes it clear to end user(s) what information the robotic system is processing and how the system makes decisions.Why involve end users? It is difficult to foresee potential use at design time. When robots meet the real world, they are bound to encounter all kinds of unforeseeable situations. Therefore, I argue that end users should be enabled to make the robot’s behavior more context-appropriate at use time. A second reason to involve end users is that end users have tacit knowledge, which is the knowledge that arises from the body and its situated interactions. This is why we need to design systems and methods that promote the inclusion of such experts. Third, I argue that when it is possible for end users to adapt the robot’s behavior, this supports their agency and self-efficacy, so that people are enabled to act in accordance with their aims and perceive themselves as capable of doing so.In the dissertation, I investigate different methods to include (representative) end users, stakeholders and domain experts in the process of determining robot behavior, in several different contexts. Care home residents and care workers are involved in discussing application scenarios of a to-be-developed technology in a co-design study. I study and develop end-user programming interfaces in manufacturing and for programming robot animations in social interaction and dance. Finally, I conduct a study on representing a robot’s knowledge base to users in an object organization scenario. The contribution of my work is a better understanding of interaction design factors and methods that enhance human agency in HRI scenarios by making it possible to impact robot behavior.