Footstep localization and force estimation through structural vibrations using the FEEL Algorithm

Abstract

Locating individuals within a space has numerous potential uses within a smart environment. Many different technologies have been explored towards this end though privacy-concerns, the need to wear a device, and/or extensive building modifications have presented challenges towards adoption. Structural vibrations caused by footsteps have been shown to overcome these challenges though current methods rely on time-of-flight variations. This paper presents the use of the Force Estimation and Event Localization (FEEL) Algorithm that utilizes lower sampling rates, less sensors, etc than time-of-flight methods towards locating persons. Improvements to FEEL’s force estimation and SDFE localization method are additionally presented with demonstrated increases in accuracy. Analysis of 1100 footsteps resulted in 98.6% localization accuracy using the improved FEEL. Ground force reactions (GRF) were estimated by FEEL and used to estimate participant body–weight-ratios (BWR). Estimated BWRs were within ranges reported in previous works for both barefoot and shoe cases.