In manufacturing environments, operators frequently need to input three-dimensional (3D) spatial positions to robotic systems to perform tasks such as grasping, assembly, and placement. Consequently, 3D pointing has become an important input modality in human-robot interaction for industrial applications. Previous studies have shown that different sensory cues delivered through augmented reality can improve operator performance in pointing tasks. This paper systematically evaluates the effects of visual, auditory, and haptic cues on operational performance during the manipulation of a Phantom Touch pointer to reach a target point across four real-virtual interaction configurations. The effects of different sensory modalities are compared using metrics including operational accuracy, task completion time, and subjective measures. The results indicate that sensory cues do not significantly improve pointing accuracy under aligned configurations, where both the pointer and target are either real or virtual. In contrast, under misaligned configurations, visual cues outperform other modalities in pointing accuracy, while haptic cues yield the shortest task completion time among the three modalities. These findings support a cue design principle for selecting appropriate sensory cues based on the interaction configuration and task objectives, providing practical guidance for AR-assisted human-robot interaction.



