Walkie-Markie: Indoor Pathway Mapping Made Easy
summary
Walkie-Markie automatically reconstructs indoor pathway maps by detecting WiFi-Marks — RSS tipping points — and fusing crowdsourced inertial-phone trajectories with a spring-relaxation graph embedding (Ataria) that uses displacement and heading. The system builds usable maps after about 5–6 walks per segment, achieves 90% node/shape accuracy within ~2 m, maximum discrepancies 3.0 m (nodes) and 2.8 m (segments), and yields localization errors around 1.7/2.9 m (50/90%), outperforming RADAR’s 2.3/5.2 m in the evaluated buildings.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 0:30Locationbased services
- 1:24Types of maps
- 1:48Problem
- 3:28Challenges
- 4:32Landmarks
- 5:59WiFi Infrastructure
- 7:01Key Idea
- 8:10Measurement Study
- 11:41Building Password Maps
- 12:25Graph Embedding
- 13:57Direction Information
- 14:38Net Force
- 15:40Results
- 15:51Simulations
- 16:17Visual Result
- 17:51Metrics
- 19:14Conclusion
- 19:39Future plans
related talks
21mEnvironmental Factors in Indoor Navigation Based on Real-World Trajectories of Blind Users
21mIndoor mapping for OpenStreetMap
57mBuilding the best maps in the world: A conversation with leaders from Mapbox and Tripadvisor
01h 34mBuild a Web Map with Open Source Tools
19mThe Map as an Information Ecology
05mMaps of My Life
19mMoving Through Country
14mTelling Traces