Please use this identifier to cite or link to this item:
|Scopus||Web of Science®||Altmetric|
|Title:||bTracked: Highly accurate field deployable real-time indoor spatial tracking for human behavior observations|
Van Nguyen, H.
|Citation:||MobiQuitous '18 Proceedings of the 15th EAI International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services, 2018, pp.1-10|
|Publisher:||Association for Computing Machinery|
|Series/Report no.:||ACM International Conference Proceeding Series|
|Conference Name:||EAI International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services (Mobiquitous) (5 Nov 2018 - 7 Nov 2018 : New York City, NY)|
|Michael Chesser, Leon Chea, Hoa Van Nguyen, Damith C Ranasinghe|
|Abstract:||Methods for accurate indoor spatial tracking remains a challenge. Low cost and power efficient Bluetooth Low Energy (BLE) beacon technology's ability to run maintenance-free for many years on a single coin cell battery provides an attractive methodology to realize accurate and low cost indoor spatial tracking. However an easy to deploy and accurate methodology still remains a problem of ongoing research interest. We propose a field deployable tracking system based on BLE beacon signals together with a particle filter based approach for online and real-time tracking of persons with a body-worn Bluetooth receiver to support fine grain human behavior observations. First, we develop the concept of generic sensor models for generalized indoor environments and build pluggable sensor models for re-use in unseen environments during deployment. Second, we exploit pose information and void constraints in our problem formulation to derive additional information about the person tracked. Third, we build the infrastructure to easily setup and operate our tracking system to support end-users to remotely track ambulating persons in real-time over a web-based interface. Fourth, we assess five different tracking methodologies together with two approaches for formulating pose information and show that our method of probabilistic multilateration including the modeling of pose leads to the best performance; a mean path estimation error of 23.5 cm in a new indoor environment.|
|Keywords:||Human motion observations; spatial tracking; particle filter; generic sensor models; Bluetooth wearable sensors; BLE beacons|
|Rights:||© 2018 Association for Computing Machinery|
|Appears in Collections:||Aurora harvest 3|
Computer Science publications
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.