PeriphAR: Fast and Accurate Real-World Object Selection with Peripheral Augmented Reality Displays

Paper Title

PeriphAR: Fast and Accurate Real-World Object Selection with Peripheral Augmented Reality Displays

Publication Info

• Topic area: Peripheral vision-based feedback for object selection in augmented reality (AR).
• Keywords: Peripheral vision, augmented reality, gaze-based interaction, monocular AR displays, color enhancement, object selection, peripheral feedback, XR, glanceable AR, human-computer interaction.

Background and Problem

• Problem / challenge: Existing AR systems for object selection rely on central vision cues, which are unsuitable for monocular, always-on AR glasses with limited field-of-view (FOV). Peripheral vision has not been sufficiently explored as a feedback channel for object selection.
• Significance: Addressing this gap can enable lightweight, low-powered AR glasses to provide intuitive, hands-free interaction while maintaining user focus on real-world tasks.
• Motivation and related work: Prior research has focused on central vision cues, adaptive interfaces, and peripheral displays for secondary information, but these approaches assume more capable AR hardware. Peripheral vision has been shown to detect motion and high-contrast changes effectively, but its application for real-world object selection in AR remains underexplored.

Solution

• Proposed approach: PeriphAR, a visualization technique that uses peripheral vision for feedback during gaze-based selection on monocular AR displays.
• Novelty:
  1. Empirical insights into effective peripheral cues for real-world object selection, enhancing accuracy and user confidence.
  2. Design and implementation of a color enhancement algorithm to improve peripheral visibility of selected targets.
  3. Evaluation of two strategies for peripheral proxy generation: snapshot-based and most-similar-color (MSC) strategies.
  4. Proof-of-concept end-to-end system for real-world object detection and selection.
• Procedure and key techniques:
  1. Simulated monocular AR display using Quest Pro with controlled experiments.
  2. Development of a color enhancement algorithm to maximize contrast between target and similar objects.
  3. Two user studies to evaluate peripheral cues and proxy generation strategies.
  4. End-to-end system implementation with real-time object detection and segmentation using YOLO11n.

Results

• Concrete findings:
  • Peripheral color cues were the most effective for selection tasks, achieving the fastest task completion time (30.4s) and the best real-world to in-display gaze ratio (6.0:1).
  • The MSC strategy reduced errors (17.7% on yellow shelves) compared to the snapshot strategy (26.2%) in cluttered environments.
  • Red targets were noticed significantly faster than green or yellow, consistent with preattentive color processing.
• Advantage over baselines:
  • The MSC strategy improved peripheral proxy distinctiveness in visually challenging scenarios, outperforming the snapshot and baseline conditions in subjective ratings (e.g., confidence, ease of noticing).
  • The color condition in Study 1 outperformed text, shape, and snapshot conditions in terms of task efficiency and user comfort.
• Experiments / evaluation:
  • Study 1 (32 participants): Tested text, color, shape, and snapshot conditions using Tetris-like virtual objects.
  • Study 2 (12 participants): Compared baseline, snapshot, and MSC strategies using virtual fruit shelves.
  • End-to-end system: Tested in real-world scenarios (e.g., vending machines, bookshelves) with average latency between 0.72s and 1.55s.
• Limitations and future work:
  • Real-world deployments require safeguards for safety-critical tasks and adaptation to lighting conditions.
  • The algorithm struggles with multi-colored objects where dominant regions are visually similar.
  • Future work includes patch-level color enhancement, neighbor-aware constraints, and context-aware parameter tuning.

Summary

PeriphAR introduces a novel approach for gaze-based object selection using peripheral vision on monocular AR displays. By leveraging color enhancement and peripheral proxies, it improves selection accuracy, reduces cognitive load, and enhances user confidence. Two user studies demonstrated the effectiveness of peripheral color cues and the MSC strategy for proxy generation. A proof-of-concept end-to-end system validated the approach in real-world scenarios, highlighting areas for refinement, such as handling multi-colored objects and adapting to lighting conditions. PeriphAR opens new possibilities for lightweight, always-on AR glasses by aligning hardware constraints with human perceptual strengths.