EchoSpot: Spotting Your Locations via Acoustic Sensing
summary
EchoSpot implements an indoor locator that uses single speaker + single microphone devices such as Amazon Echo Edifier R1280DB and Logitech z200 to emit FMCW chirps 18–23 kHz and capture reflections. It computes normalized cross correlation to extract time-of-flights, isolates body versus wall echoes by treating the wall as a virtual sender, triangulates 2D position from body and wall ToFs, and smooths output with a Kalman filter. Implemented in real homes it achieves mean errors of 4.1 cm@1m, 9.2 cm@2m, 13.1 cm@3m, 17.9 cm@4m, 22.2 cm@5m, median static 12.4 cm and average tracking error 21.9 cm.
AI-written summary · about