VisiMark: Characterizing and Augmenting Landmarks for People with Low Vision in Augmented Reality to Support Indoor Navigation

Research Background and Issues

• What problems or challenges did the authors identify?

  • Navigation is a complex task for people with low vision (PLV), who have partial visual capabilities. They rely on landmarks as navigation references, but due to their visual limitations, identifying certain landmarks can be challenging, such as those with low contrast, small size, or located outside their field of view.
  • While previous studies have explored how blind and sighted individuals use landmarks, PLV's visual abilities fall between these two groups. Their unique needs for landmark selection and enhancement have been largely understudied.

• Why is this issue important?

  • Landmarks play a critical role in spatial navigation, helping users build mental models, self-orient, and ensure safety. Understanding how PLV use landmarks during navigation and how to effectively enhance these landmarks is crucial for improving the convenience, safety, and independence of indoor navigation.

• Research Motivation and Related Work

  • The authors aim to address two core questions: ① How do PLV select, perceive, and use landmarks for navigation and mental model construction? ② What types of enhancement techniques can effectively support PLV in recognizing landmarks?
  • Given that augmented reality (AR) technology can provide immersive, multimodal visual feedback, the authors seek to explore how AR can be leveraged to enhance PLV's ability to recognize landmarks.

Solutions

• What methods or solutions did the authors propose?

  • The authors designed VisiMark, a head-mounted AR-based interface for landmark enhancement, aimed at improving PLV's landmark perception. VisiMark includes two main features:
    1. Signboards: Providing previews of spatial structures and upcoming landmarks at corridor intersections.
    2. In-situ Labels: Enhancing landmarks in the physical environment with icons and text.

• What is innovative about this solution?

  • VisiMark not only displays landmarks but also provides an overview of spatial structures through AR, allowing PLV to anticipate landmarks that may not be directly visible to the naked eye.
  • The proposed landmark enhancement approach focuses not only on visual salience but also on cognitively important yet visually challenging landmarks, helping PLV move beyond reliance on traditionally salient landmarks.

• What are the implementation steps and key technologies used?

  • The authors first conducted user research to gain insights into how PLV select and use landmarks, which informed the design of the VisiMark interface.
  • In implementation, VisiMark uses the Microsoft HoloLens 2 headset to pre-scan environments and calibrate landmark positions, with AR content developed using Unity.
  • The system offers various customization options (e.g., font size and color) to accommodate users' visual preferences. Additionally, a "Wizard of Oz" method was employed to adjust system parameters for interface optimization.

Research Outcomes

• What specific outcomes were achieved?

  • VisiMark significantly enhanced PLV's ability to perceive landmarks, enabling users to better identify previously hard-to-detect but critical landmarks for navigation and safety (e.g., elevators and restrooms hidden in walls).
  • The system effectively shifted PLV's landmark selection from solely visually salient objects to cognitively important and more meaningful landmarks.
  • Experiments demonstrated that PLV using VisiMark performed better in navigation and path reproduction tasks, such as improving the accuracy of landmark memory in mental map construction tasks.

• What advantages does it have compared to existing solutions?

  • Compared to traditional landmark enhancement tools, VisiMark not only improves the visual appearance of landmarks but also provides an overview of spatial structures, helping PLV recognize both global and local landmarks.
  • The system offers detailed information through icons and text labels, which aids users in memorizing key features of landmarks more easily.

• What were the experimental or evaluation results?

  • Users achieved higher accuracy in reproducing navigation paths and significantly improved landmark memory accuracy when using VisiMark.
  • Subjective feedback indicated that users found the system effective and easy to learn, with the majority reporting that VisiMark did not cause visual interference.

• Limitations and Future Directions

  • The authors noted that the system relies on pre-scanned environments, limiting its applicability in real-time dynamic settings. Future work should incorporate real-time AI recognition technologies to support landmark enhancement in dynamic environments.
  • The current study focuses only on indoor environments, and further research is needed to address complex outdoor scenarios.
  • The types of landmark enhancements are still limited; future work could expand to include more interactive components (e.g., door handles, floor obstacles) to meet users' safety needs.
  • The enhancement experience requires more flexible customization options to dynamically adjust the position and range of enhanced elements based on users' functional visual areas.

This study makes significant contributions to the field of landmark enhancement for PLV, offering practical advancements in navigation technology for individuals with low vision. Through in-depth analysis of landmark classification and enhancement methods, it provides valuable guidance for the design of future navigation systems.