Human-Robot Interaction Experiment: Minor Changes; Significant Differences

Abstract

Designing successful human-robot interaction (HRI) experiments is a complex and challenging task for researchers, with no comprehensive guides or established best practices available to navigate this intricate process. The field of HRI often involves trial and error, resulting in numerous unsuccessful experiments or witnessing colleagues’ struggles in testing their hypotheses. In this paper, we outline the rigorous two-and-a-half-year journey of our team in developing a successful human-robot trust experiment. We designed a 2D search task in which a human and a robot collaborated to achieve a higher team score. The only requirement for effective collaboration was trust, which was challenging to maintain given the robot’s varying reliability. In this paper, we present three of our most significant experiment designs, discussing the outcomes of each, the reasons behind the initial failures, and the modifications implemented at each stage. The designed game can be conducted remotely or in person, with optional supplementary measures. Our insights aim to contribute to the body of knowledge in HRI experiment design, offering practical guidance and lessons learned for future research endeavors.