Estimating Movement Smoothness from Inertial Measurement Units
Author: Melendez-Calderon, A., Shirota, C., and Balasubramaian, S. 2021. Estimating Movement Smoothness From Inertial Measurement Units, Frontiers in Bioengineering
and Biotechnology. 2021, 14 January. https://www.frontiersin.org/articles/10.3389/fbioe.2020.558771/full
Published: 2 June 2021
“Movement smoothness is a fairly intuitive concept: most people can easily distinguish the well-coordinated, free-flowing movements of a professional ballerina from those of a beginner.”
Movement smoothness is a fairly intuitive concept: most people can easily distinguish the well-coordinated, free-flowing movements of a professional ballerina from those of a beginner.
Aligned with our intuition, smoother movements are related to higher skill, as well as recovery following neurological injury. Despite the almost obvious intuitive concept of movement smoothness, (1) there are few methods to quantify it, that is, to assign a number indicating the “amount of smoothness” of a movement, (2) these methods are not yet standardised, and (3) smoothness values may differ depending on the movement sensor technology used.
Movement sensors, such as those commonly found in phones and fitness trackers, are becoming more and more common in everyday electronics, and provide an opportunity to more easily measure movement smoothness.
These sensors, called inertial measurement units (IMUs), ‘detect movement’ (by measuring translational acceleration and rotational velocity). One challenge with these measurements is that they cannot be directly used with methods that measure movement smoothness.
This paper takes a deep dive into the basic physics of inertial measurement units and movement smoothness to: (a) bring to light the various theoretical and practical issues in using existing measures of movement smoothness on data from inertial measurement units, and (b) propose an alternative method to quantify smoothness directly using these sensors.
In summary: movement smoothness can be reliably calculated from inertial measurement units, but *much care* needs to be taken when doing so. Smoothness measures are affected by gravity, insufficient information measured by IMUs, noise, and more – which are detailed in the paper.
We hope that this paper will promote more consistent and widespread use of the appropriate methods to quantify movement smoothness to benefit the study of human movement and rehabilitation.
View a PDF Version: Movement Smoothness
The Hopkins Centre
Menzies Health Institute QLD
0478 709 990