Egocentric depth judgments in optical, see-through augmented reality

Area of Science:

• Human-Computer Interaction
• Computer Vision
• Perception Psychology

Background:

• Optical see-through augmented reality (AR) systems face challenges in accurately representing spatial layout and depth.
• Ensuring users perceive virtual graphics in consistent spatial relationships with real-world objects is crucial for AR development.
• AR enables novel perceptual capabilities, such as 'x-ray vision,' requiring accurate depth perception for virtual objects behind real-world occlusions.

Purpose of the Study:

• To review and discuss protocols for measuring egocentric depth judgments in virtual and augmented environments.
• To investigate depth underestimation commonly observed in virtual environments.
• To experimentally measure egocentric depth judgments in AR across different distances and conditions.

Main Methods:

• Experiment I utilized a perceptual matching protocol to assess AR depth judgments from 5 to 45 meters, examining visual field location, 'x-ray vision,' and practice effects.
• Experiment II employed blind walking and verbal reports to measure AR depth judgments between 3 and 7 meters, comparing real, virtual, and mixed objects.
• Protocols reviewed include those for measuring egocentric depth in both virtual reality (VR) and AR settings.

Main Results:

• Experiment I indicated a bias shift, with AR depth judgments switching from underestimation to overestimation around 23 meters.
• The 'x-ray vision' condition in Experiment I was found to significantly increase task difficulty.
• Experiment II results suggest egocentric depth in AR is underestimated at near distances (3-7m), though less so than in typical VR environments.

Conclusions:

• AR depth perception exhibits complex behavior, including a distance-dependent shift in judgment bias.
• Factors such as restricted field-of-view and limited scanning strategies may contribute to depth underestimation in AR at near distances.
• Understanding and mitigating depth perception errors are essential for developing effective and intuitive AR experiences.