Synchronous Wireless Body Sensor Network Enabling Human Body Pose Estimation

• Published in: IEEE access Vol. 7; pp. 49341 - 49351
• Main Authors: Laurijssen, Dennis, Saeys, Wim, Truijen, Steven, Daems, Walter, Steckel, Jan
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Base stations; Biological system modeling; Body area networks; Body sensor networks; distributed embedded systems; Elevation angle; Estimation; Hardware; Human body; Infrared detectors; Infrared lasers; Integrated circuits; Maximum likelihood estimators; motion capture; Nodes; Photodiodes; Pose estimation; sensor arrays; Sensors; Solution space; Wireless communication; Wireless networks; Wireless sensor networks
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2019.2910636

In this paper, a novel human body pose estimation system is introduced, which features a synchronous network of wirelessly connected sensor nodes, thus forming a wireless body sensor network (WBSN). This measurement architecture for establishing human body pose estimations works in combination with the HTC Vive base stations and a Wi-Fi router. The former emits infrared pulses and laser sweeps in the horizontal and vertical plane captured by the sensor nodes' photodiodes. The infrared data can be converted into an azimuth and elevation angle for each photodiode relative to the emitting HTC Vive device. To estimate the human body poses, the sensor data are fed into a probabilistic non-linear maximum likelihood estimator combined with a parametric human body model. Using a parametric human body model for estimating the poses proves to be more resilient to sensor noise compared to estimating the Six Degrees of Freedom (6DoF) of every sensor node individually. The solution space can be constrained to the parameters of the users body model that relates to a priori information on the subject. The combination of the proposed hardware sensor network, its synchronous sensor data, and processing algorithms yields a cost-efficient human body pose estimation system.