{"id":19552,"date":"2025-11-27T09:37:26","date_gmt":"2025-11-27T01:37:26","guid":{"rendered":"https:\/\/cde.nus.edu.sg\/edic\/?page_id=19552"},"modified":"2025-12-12T16:23:06","modified_gmt":"2025-12-12T08:23:06","slug":"presence-location-intelligence","status":"publish","type":"page","link":"https:\/\/cde.nus.edu.sg\/edic\/projects\/innovating-for-smarter-living\/presence-location-intelligence\/","title":{"rendered":"Presence: location intelligence"},"content":{"rendered":"\n

\n\t\tPresence: location intelligence\n\t<\/h2>\n\t

Bang & Olufsen (B&O), a Danish luxury audio company, envisions future sound systems that respond intuitively to a listener’s position – creating seamless, spatially adaptive audio experiences across rooms. Achieving this requires precise, room-scale indoor localisation.<\/p>\n

This project develops an Ultra-Wideband (UWB)-based middleware that bridges raw radio-frequency ranging data to a usable, high-resolution spatial grid. The middleware fuses data from multiple distributed anchors to determine a user’s live position. Our design follows a three-layer architecture:<\/p>\n

    \n
  1. Edge layer<\/strong>: Multiple NXP Type-2BP UWB modules collect Time-of-Flight and Angle-of-Arrival data from an iPhone acting as a UWB transmitter.<\/li>\n
  2. Communication layer<\/strong>: A distributed MQTT Pub-Sub framework allows each anchor to publish its data over Wi-Fi to a central broker, ensuring scalability and resilience across rooms.<\/li>\n
  3. Processing layer<\/strong>: A Pose Graph Optimisation (PGO) algorithm fuses all incoming measurements into a globally consistent position estimate. Outlier rejection and a sliding-window filter reduce noise and reject erroneous readings in real time.<\/li>\n<\/ol>\n

    Across our controlled test rig, the middleware consistently improved position accuracy by ~32% over the worst-anchor baseline,\u00a0 while rejecting roughly 10% of noisy data before fusion. The system not only reduced mean error by 3-4x, but also improved reliability. This demonstrates that even imperfect UWB readings can, through sensor fusion, yield stable, high-fidelity location data suitable for\u00a0 responsive multi-room sound experiences.<\/p>\n

    \"\"<\/p>\n\t

    Project Team<\/h3>\n

    Students:<\/h4>\n