User-Centered Design and Evaluation of a Mobile Shopping Robot

This paper describes the user-centered design and evaluation process of a humanoid mobile shopping robot named TOOMAS that assists customers of home improvement stores. Three separate empirical field studies addressing the robot’s usability (according to ISO 9241-11) and acceptability (intention to use) are presented involving N = 343 test persons altogether. The first formative evaluation study (N = 210) addresses the usability of the robot’s article search system. It is demonstrated how several usability problems could be identified and eliminated, leading to significantly more successful article searches. The second formative evaluation study (N = 39) addresses the robot’s adaptation to its specific task and role in the home improvement store. Embodiment, mobility, voice output, and social behavior were anaN. Doering · S. Poeschl (B) Institute of Media and Communication Science, Media Psychology and Media Design Group, Technische Universität Ilmenau, P.O.B. 10 05 65, 98684 Ilmenau, Germany e-mail: [email protected] N. Doering e-mail: [email protected] H.-M. Gross Neuroinformatics and Cognitive Robotics Lab, Department of Computer Science and Automation, Technische Universität Ilmenau, P.O.B. 10 05 65, 98684 Ilmenau, Germany A. Bley · C. Martin MetraLabs Neue Technologien und Systeme GmbH, Am Vogelherd 22, 98693 Ilmenau, Germany H.-J. Boehme Faculty of Information Technology/Mathematics, University of Applied Sciences Dresden, P.O.B. 12 07 01, 01008 Dresden, Germany lyzed and adapted to user requirements. The third summative evaluation study (N = 94) experimentally tested robotassisted shopping against conventional shopping regarding core usability criteria (effectiveness, efficiency, and satisfaction). It reveals that robot-assisted shopping was as effective and satisfactory as conventional shopping. Still, at the current state of technology shopping with robot assistance was slightly slower and therefore less efficient. For all three studies the effects of users’ gender, age, educational background and computer skills levels on robot usability and acceptability are presented and discussed.