Posture Detection Using Sounds and Temperature: LMS-Based Approach to Enable Sensory Substitution

• Published in: IEEE transactions on instrumentation and measurement Vol. 67; no. 7; pp. 1543 - 1554
• Main Authors: Russell, L., Goubran, R., Kwamena, F.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.07.2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Acoustic; Acoustic noise; Acoustics; adaptive signal processing; Algorithms; Biomedical monitoring; Change detection; Cushions; Furniture; Intelligent sensors; Internet of Things (IoT); least mean square (LMS); low cost; Mats; Measurement methods; Monitoring; normalized least mean square (NLMS); Occupancy; Parameters; Posture; pressure; Pressure measurement; Pressure sensors; Security; Sensor arrays; Sensors; smart environment; Substitutes; Temperature measurement; temperature sensor; Temperature sensors; unobtrusive sensors
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2018.2795158

Sensors are used to determine a variety of health and security parameters. Pressure mats and camera analysis are frequently used to determine a person's chair posture. Measurement of parameters can be derived from a variety of different sensor types, which may already be present in an environment. This paper presents the use of the concept of "sensory substitution," where a sensor designed to measure Quantity X is used to measure Quantity Y. The concept is used to enable alternative sensing techniques for medical and security applications, such as determining chair occupancy, postural shift times, and even current postural state, without using cameras or pressure sensors. Specifically, the postural state of a subject is determined in a laboratory study using various arrays of both temperature sensors and acoustic sensors. This method can be used standalone, can augment other sensors, or can validate data from pressure or visual-based systems. These alternative sensors could also be used for other aspects of smart infrastructure. The system was tested with different types and styles of furniture, including chairs, armrests, cushions, and beds. A least mean square algorithm was deployed to remove noises enabling acoustic posture detection. Using a thresholding algorithm, postural changes are then identified using audio. In a controlled environment, over 90% of postural change events were detected. Thus, the paper shows two alternative instrumentation and measurement methods to determine occupancy, postural change timings, and even posture states of a person in a chair.