How People’s Perception on Degree of Control Influences Human-Robot Interaction

Abstract

Automated products that seem to be more sophisticated every day are invading the market. Gmail provides suggestions for emails responses and can even track important dates through emails and send a notification about it without the user's permission. As robot companions are just slowly starting to be available to the public, one must wonder, do people expect robots to have the same technology advancements as other technology tools such as smart phones? Is it really what people want? Some early research on control has been made in the Human Computer Interaction community by Shneiderman & Maes (1997) to discover how much control the user is ready to give up to an intelligent agent. This PhD does the same type of investigations for domestic robots by focussing on perception of control in Human-Robot Interaction (HRI). To be able to conduct such an investigation, the user's perception of control is measured through the robot's level of autonomy. As this thesis will show, little research has been done in this area for domestic robot companions. After a first exploratory study was conducted to gain a better understanding of perception of control related to the user's preferred level of autonomy of the robot for a simple task (cleaning), three questionnaire studies have investigated what makes a task high critical or low critical, physical or cognitive. The results could then be used to design a full live investigation on how the level of criticality of a task influence the user's preference of the robot's level of autonomy. The results of this thesis show that in general people want robots to be more autonomous but they still want to have control over the robot for most tasks. People prefer to give instructions to the robot when a cognitive task is performed regardless of the criticality of the task, and for a low critical physical task that is entertainment-based. However, for a high critical physical task, the user prefers the robot to be fully autonomous even if they feel they have less control over the robot. This is explained by the way participants perceived the performance of the task. When the robot was fully autonomous, they felt the task was done faster and smoother than when they had to continuously provide instructions to the robot.