Augmented Reality as a Proxy for Human–Robot Interaction Research: Comparing Subjective and Objective Measures

Abstract

Augmented reality (AR) enables simulating robots in Human-Robot Interaction (HRI) research, offering a cheaper, more accessible, and flexible alternative to complex, resource-intensive physical setups. However, the comparability of AR and physical HRI results remains unclear. In this paper, we address this question by comparing subjective workload and engagement ratings, and objective task performance in AR-simulated and physical robot settings under fatigue. Participants (N = 47) completed a collaborative assembly-and-handover task with either an AR-simulated or a physical robot. Subjective workload and engagement were comparable between environments and, for several measures (namely mental demand and perceived effort), showed practical equivalence. Objective task performance differed for some measures: reaction times and movement durations were longer in AR, whereas more errors occurred in the physical condition, while assembly duration showed equivalent results. We discuss these findings with respect to opportunities and limits to the validity of AR-simulated HRI research.