DreamWalker: Substituting Real-World Walking Experiences with a Virtual Reality
summary
System that fuses dual-band GPS, inside-out tracking, and dual RGBD cameras plus GPU processing to position users to centimeter-scale, detect static/ad-hoc/dynamic obstacles via height/color/ground maps and YOLO pedestrian detection, and redirect real walking onto matched pre-authored VR paths. Built and fielded across Microsoft campus: 8 participants walked ~600 m routes (9–15 minutes, described as 15-minute route), users prefer DreamWalker and walk faster though raw RGBD feels safer.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:33Walkthrough
- 2:30Related Walks
- 4:30Path Training
- 6:00Environment Sensing
- 8:38RealTime Tracking
- 10:55Mapping
- 13:40Evaluation
- 15:53Questions
related talks
09mVR-STEP: Walking-in-Place using Inertial Sensing for Hands Free Navigation in Mobile VR Environments
17mIncreasing Walking in VR using Redirected Teleportation
12mDo You See What I See? Bring Live Pedestrians into an Outdoor Collaborative Mixed Reality Experience
09mVRGaitAnalytics: Visualizing Dual Task Cost for VR Gait Assessment
10mPerceived realism of pedestrian crowds trajectories in VR
15mOutstanding: A Multi-Perspective Travel Approach for Virtual Reality Games
46mScience based Design Decision Making for Virtual Reality
20mOptical Marionette: Graphical Manipulation of Human's Walking Direction