Large Network UWB Localization: Algorithms and Implementation
summary
Locate3D is a UWB-based system that uses antenna-array joint range and angle measurements to make each peer-to-peer exchange four times more informative, cutting measurement count ~4× and latency up to 75%. It selects quality links via a weighted minimum spanning tree, splits graphs into rigid subclusters using a rigidity matrix eigenanalysis, and aligns local frames for full 6-DOF poses. Prototype and simulator results: 0.86 m median error for a 32-node multi-floor deployment (0 anchors) and 12.09 m for 100,000 nodes with 15 anchors.
AI-written summary · about
chapters (14)
- 0:00Large-scale UWB localization intro
- 1:19Why current systems don't scale
- 2:22Graph-based localization with MDS
- 3:05Three scaling challenges
- 4:54Locate3D core ideas
- 5:37Joint range and angle optimization
- 6:41Optimal edge selection with spanning trees
- 7:46Handling partial rigidity
- 9:11Rigidity matrix analysis
- 10:14Reference frame alignment
- 11:20Prototype testing results
- 12:04Building-scale 32-node test
- 12:46City-scale simulation accuracy
- 13:28Locate3D's four contributions
related talks
06mULoc: Low-Power, Scalable and cm-Accurate UWB-Tag Localization and Tracking for Indoor Applications
21mUWB With Arduino: Unlocking Precise Positioning for IoT
18mLocAP: Autonomous Millimeter Accurate Mapping of WiFi Infrastructure
34mThe Easy and the Not-so-Easy of Indoor Tracking and Localization Systems
17mRF-Chord: Towards Deployable RFID Localization System for Logistic Networks
24mPinPoint: Localizing Interfering Radios
21mTracko: Ad-hoc Mobile 3D Tracking Using Bluetooth Low Energy and Inaudible Signals ...
19mIndoor Position with IVY5661 96Boards