Optimizing wearable motion tracking by assessing sagittal joint angle accuracy with minimal sensor use

Abstract

Introduction

Wearable motion tracking technology often focuses on reducing the number of sensors to simplify design and lower costs. Research has shown that single IMUs can reconstruct leg kinematics (Gholami et al., 2020; Hossain et al., 2022; Lim et al., 2020) and ground reaction forces (Jiang et al., 2020) effectively. Additionally, model-based methods have demonstrated the feasibility of using fewer gyroscopes to estimate stride length and motion range in healthy individuals and patients with coxarthritis (Salarian et al., 2013). In this study, we aim to assess the precision of sagittal joint angle estimations using strain sensors while minimizing sensor count.

Methods

We conducted a study with ten participants based on our previous work that involved collecting single-leg treadmill running data to monitor lower limb joint angles with piezoresistive strain sensors. Subjects ran on an instrumented treadmill at 8-10 km/h, wearing athletic pants embedded with nine strain sensors located on the hip, knee, and ankle. Optical motion capture provided reference kinematics. Our prior research achieved less than 1.5° error in the sagittal plane using a machine-learning approach. The current study explores the extent to which sensor reduction is possible without meaningful loss of accuracy. Three evaluation measures were used for assessment: Pearson correlation, dynamic time warping, and root-mean-squared error.

Results

The results from our correlation analysis will be used to develop a model that optimally balances between accuracy and minimizing the number of sensors. This has practical implications in sports science, where athletes could benefit from less intrusive and more comfortable performance monitoring, and in healthcare, for remote monitoring of patients with mobility issues.

References

Gholami, M., Napier, C., & Menon, C. (2020). Estimating lower extremity running gait kinematics with a single accelerometer: A deep learning approach. Sensors, 20(10), Article 2939. https://doi.org/10.3390/s20102939

Hossain, M. S., Bin, Dranetz, J., Choi, H., & Guo, Z. (2022). DeepBBWAE-Net: A CNN-RNN based deep superlearner for estimating lower extremity sagittal plane joint kinematics using shoe-mounted IMU sensors in daily living. IEEE Journal of Biomedical and Health Informatics, 26(8), 3906-3917. https://doi.org/10.1109/jbhi.2022.3165383

Jiang, X., Napier, C., Hannigan, B., Eng, J. J., & Menon, C. (2020). Estimating vertical ground reaction force during walking using a single inertial sensor. Sensors, 20(15), Article 4345. https://doi.org/10.3390/s20154345

Lim, H., Kim, B., & Park, S. (2020). Prediction of lower limb kinetics and kinematics during walking by a single IMU on the lower back using machine learning. Sensors, 20(1), Article 130. https://doi.org/10.3390/s20010130

Salarian, A., Burkhard, P. R., Vingerhoets, F. J. G., Jolles, B. M., & Aminian, K. (2013). A novel approach to reducing number of sensing units for wearable gait analysis systems. IEEE Transactions on Biomedical Engineering, 60(1), 72–77. https://doi.org/10.1109/TBME.2012.2223465