Replayable Augmented Reality Visualization for Robot Fault Diagnosis: A Comparative Study

Abstract Efficient fault diagnosis in autonomous robotic systems is essential for minimizing downtime. This study compares the effectiveness of an Augmented Reality (AR) interface and sensor data replay (featuring a 15-second loop before the fault) in diagnosing common robot faults. In a user study, 24 participants experienced a series of eight staged robot fault scenarios. A tablet-based interface, presenting identical information, was favoured over AR due to its effectiveness and ease of use. Participant feedback highlighted the limitations of the AR interface including the low field of view and blurriness, suggesting potential improvements in future AR headset iterations. While a preference for replayed data emerged, it was not supported uniformly, warranting further research. This study advances the exploration of Augmented Reality in human-robot interaction, emphasizing the crucial role of user-friendly interfaces for efficient robot fault diagnosis.