A Sensor Placement Benchmarking Method With Principal Component Analysis

Fengyu Zhong, Liang He, Amy Zavatsky, Perla Maiolino

Abstract: Foot plantar pressure measurements are valuable for diverse applications in clinical and biomedical studies. In-shoe devices have recently become popular and are considered effective tools for acquiring foot plantar pressure data, thanks to their versatility. However, these devices are subject to low measurement resolution because only a limited number of sensors can be installed due to considerations of cost and comfort. Optimizing the sensor layout could be an effective solution to enhance the measurement quality. Contemporary devices rely either on an understanding of anatomy and practical experience or expensive data-driven approaches to configure the sensor layouts. In this work, we present a sensor placement benchmarking method that bridges the gap between the two aforementioned categories of methods and helps determine sensor layouts efficiently. The benchmarking method utilizes an open-access dataset to evaluate a sensor layout from multiple perspectives, including the accuracy of center of pressure (CoP) estimation, prediction rate, sensor count, and the physical insights generated from conducting principal component analysis (PCA) on the dataset. Subsequently, the study demonstrates the method’s functionality by applying it to the anatomy-based sensor layouts of two existing devices. Optimal sensor placements are identified, and general guidelines for sensor placement are proposed.

External IDs:doi:10.1109/tim.2023.3306817