Multi-Modal Sensing System Integrating a Pressure Sensor With Line-Scan LiDAR for Intelligent Sitting Posture Classification
IEEE Access · 2026 · 14 · 84719-84731
DOI: 10.1109/access.2026.3699682 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Daehyeon Ji und 3 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Korea Planning and Evaluation Institute of Industrial Technology, funded by the Ministry of the Interior and Safety (MOIS), South Korea, Chonnam National University (Smart Plant Reliability Center), grant funded by the Ministry of Education.
Abstract (Originalsprache)
Prolonged flexed sitting postures can result in spinal misalignment and musculoskeletal disorders. To enable effective monitoring, a multi-modal sensing system was developed by integrating a flexible pressure sensor for lower-body assessment with a one-dimensional (1D) light detection and ranging (LiDAR) for upper-body evaluation. The low-cost LiDAR provided line-scan profiles to measure spinal curvature, allowing accurate classification of upper-body postures that were difficult to distinguish in previous studies. Six fusion strategies (early fusion (EF), intermediate fusion (IF), early–intermediate fusion (EIF), max confidence fusion (MCF), weighted average fusion (WAF), and conditional fusion (CF) were combined with four classifiers (artificial neural network (ANN), support vector machine (SVM), 1D convolutional neural network (1D-CNN), and 2D-CNN), resulting in 24 multi-modal models capable of classifying 12 sitting postures. Among these, through the systematic analysis of performance metrics and prediction time, the EF with 1DCNN (EF-1DCNN) was found to be the most balanced model, demonstrating both a high classification accuracy (92.74%) and a competitive prediction time (0.2 s), outperforming both single-modal methods (pressure sensor–based: 77.08%, LiDAR-based: 33.26%) and the other fusion strategies. Unlike earlier pressure-sensor-based systems that achieved only 68.74–81.80% accuracy in detecting slouching, the proposed system nearly eliminated these errors. These results demonstrate the potential of the proposed sensor for applications requiring rapid and precise posture recognition.
Förderung
- Korea Planning and Evaluation Institute of Industrial Technology, funded by the Ministry of the Interior and Safety (MOIS), South Korea
- Chonnam National University (Smart Plant Reliability Center), grant funded by the Ministry of Education
Zitieren
APA
Ji, D., Choe, Y., Park, C., & Park, Y. (2026). Multi-Modal Sensing System Integrating a Pressure Sensor With Line-Scan LiDAR for Intelligent Sitting Posture Classification. IEEE Access, 14, 84719-84731. https://doi.org/10.1109/access.2026.3699682
Vancouver
Ji D, Choe Y, Park C, Park Y. Multi-Modal Sensing System Integrating a Pressure Sensor With Line-Scan LiDAR for Intelligent Sitting Posture Classification. IEEE Access. 2026;14:84719-84731. doi:10.1109/access.2026.3699682
BibTeX
@article{Ji2026,
author = {Ji, Daehyeon and Choe, Yunseok and Park, Chang-Ju and Park, Yangkyu},
title = {Multi-Modal Sensing System Integrating a Pressure Sensor With Line-Scan LiDAR for Intelligent Sitting Posture Classification},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {84719--84731},
doi = {10.1109/access.2026.3699682}
}
EUVIMED