Exploration of Radar-based Obstacle Visualizations to Support Safety and Presence in Camera-Free Outdoor VR

TL;DR

This work addresses safety and presence in camera-free outdoor VR by leveraging a privacy-preserving mmWave radar sensing pipeline (WaveWalkerClone) that fuses GPS–IMU data to detect near-by obstacles. It validates system feasibility under outdoor lighting and compares three radar visualization strategies—diegetic alien avatars, non-diegetic human avatars, and abstract point clouds—through two studies. All visualization modes supported safe navigation, but each entailed distinct trade-offs in effort, frustration, and preference, underscoring the need for adaptable or hybrid designs. The findings offer design considerations that balance safety, immersion, and bystander privacy, advancing outdoor VR toward practical, private, and coherent mixed-reality experiences.

Abstract

Outdoor virtual reality (VR) places users in dynamic physical environments where they must remain aware of real-world obstacles, including static structures and moving bystanders, while immersed in a virtual scene. This dual demand introduces challenges for both user safety and presence. Millimeter-wave (mmWave) radar offers a privacy-preserving alternative to camera-based sensing by detecting obstacles without capturing identifiable visual imagery, yet effective methods for communicating its sparse spatial information to users remain underexplored. In this work, we developed and validated WaveWalkerClone, a reproduction of the WaveWalker system, to establish reliable radar- and GPS-IMU-based sensing under varied outdoor lighting conditions. Building on this feasibility validation, we conducted a user study (n=18) comparing three visualization techniques for radar-detected obstacles : (1) diegetic alien avatars that visually embed obstacles within the virtual narrative, (2) non-diegetic human avatars represented obstacles as humans inconsistent with the virtual narrative, and (3) abstract point clouds centered around the obstacles conveying spatial data without anthropomorphic or narrative associations. Our results show that all three approaches supported user safety and situational awareness, but yielded distinct trade-offs in perceived effort, frustration, and user preference. Qualitative feedback further revealed divergent user responses across conditions, highlighting the limitations of a one-size-fits-all approach. We conclude with design considerations for obstacle visualization in outdoor VR systems that seek to balance immersion, safety, and bystander privacy.

Figures (6)

• Figure 1: A user wearing the WaveWalkerClone outdoor VR system which comprises of a Pixel 8 smartphone which provides GPS and IMU data. A Jetson Nano fuses the GPU-IMU data, processes TI mmWave radar data, and streams output to the Meta Quest 3 headset.
• Figure 2: In our visualization study, participants explored an extraterrestrial-themed VR environment featuring real-world bystanders rendered using three visualization strategies: (a) diegetic alien avatars, (b) non-diegetic human avatars, and (c) abstract point clouds. Bystander representations as seen in VR by the study participant are outlined here with bounding boxes for clarity.
• Figure 3: Aerial view of the campus quad used as the basis for the outdoor VR environment in the city theme used in Study I. In the preliminary system evaluation, participants walked approximately 500 m around the full quad (white dashed line). In the visualization study, participants followed a shorter 200m route along one side (yellow dashed line), beginning at the circle and ending at the arrow tip.
• Figure 4: This figure presents the results from our preliminary system evaluation (Study I). The plots show (a) scores for spatial presence, realism and involvement scales of the IPQ, (b) ratings for perceived safety, enjoyment, and comfort, (c) perceived safety across different lighting conditions, and (d) a comparison of walking times between the with-VR and without-VR conditions.
• Figure 5: Participant preferences for the three visualization types are illustrated. The Diegetic Alien Avatar was selected as the first choice by 9 participants, while the Non-Diegetic Human Avatar was ranked second by 10 participants. The Abstract Point Cloud showed a more balanced distribution across rankings, with 7 participants identifying it as their least preferred option.