# Assessing real-world movements using consumer-grade wearable devices: Measuring segment orientations and movement quality

## TL;DR

This study evaluates how well consumer-grade wearable devices can measure human movement quality in real-world settings.

## Contribution

The study identifies optimal sensor placement and highlights the potential of consumer-grade wearables for real-world biomechanical assessments.

## Key findings

- Chest sensors showed the highest accuracy with mean RMSE less than 9.0° across all movements.
- Wrist sensors had considerable variability, while ankle and thigh sensors showed mixed results ranging from 2.0° to 40.0°.
- Yaw angles consistently had higher discrepancies compared to pitch and roll angles.

## Abstract

In recent years, there has been growing interest regarding the impact of human movement quality on health. However, assessing movement quality outside of laboratories or clinics remains challenging. This study aimed to evaluate the capabilities of consumer-grade wearables to assess movement quality and to consider optimal sensor locations. Twenty-two participants wore Polar Verity Sense magnetic, angular rate, and gravity (MARG) sensors on their chest and both wrists, thighs, and ankles, while performing repeated bodyweight movements. The Madgwick sensor-fusion algorithm was utilized to obtain three-dimensional orientations. Concurrent validity, quantified using the root-mean-square-error (RMSE), was established against a Vicon optical motion capture system following time-synchronization and coordinate-system alignment. The chest sensors demonstrated the highest accuracies overall, with mean RMSE (

) less than 9.0° across all movements. In contrast, the wrist sensors varied considerably (

). Ankle and thigh sensors yielded mixed results, with the

ranging from 2.0° to 40.0°. Notably, yaw angles consistently demonstrated higher discrepancies overall, while pitch and roll were relatively more stable. This study highlights the potential of consumer-grade MARG sensors to increase the real-world applicability and accessibility of complex biomechanical models. It also accentuates the requirement for strategic sensor placement and refined calibration and postprocessing methods to ensure accuracy.