Replication Data for Estimating Lower Limb Kinematics using a Reduced Wearable Sensor Count.

<p>For details on using the data and on available helper codes, refer to <a href="https://gait-tech.github.io/gaittoolbox/">https://gait-tech.github.io/gaittoolbox/</a>.</p>

<p>The experiment had nine healthy subjects (7 men and 2 women, weight 63.0 ± 6.8 kg, height 1.70 ± 0.06 m, age 24.6 ± 3.9 years old). Our system was compared to two benchmark systems, namely the Vicon and Xsens systems. The Vicon Vantage system consisted of eight cameras covering approximately 4 x 4 m^2 capture area with millimetre accuracy. Vicon data were captured at 100 Hz and processed using Nexus 2.7 software. The Xsens Awinda system consisted of seven MTx units (IMUs). Xsens data were captured at 100 Hz using MT Manager 4.8 and processed using MVN
Studio 4.4 software. The Vicon and Xsens recordings were synchronized by having the Xsens Awinda station send a trigger pulse to the Vicon system at the start and stop event of each recording. Each subject had reflective Vicon markers placed according to the Helen-Hayes 16 marker set, seven MTx units attached to the pelvis, thighs, shanks, and feet according to standard Xsens sensor placement, and two MTx units attached near the ankles. </p>

<p>Each subject performed the movements listed in the Table below twice (i.e., two trials). The subjects stood still before and after each trial for ten seconds. The experiment was approved by the Human Research Ethics Board of the University of New South Wales (UNSW) with approval number HC180413.</p>

<p>TABLE I: TYPES OF MOVEMENTS DONE IN THE VALIDATION EXPERIMENT Duration (s)</p>
<ul>
<li> Static: Stand still (~10s) </li>
<li> Walk: Walk straight and back (~30s) </li>
<li> Figure of eight: Walk in figures of eight (~60s) </li>
<li> Zig-zag: Walk zigzag (~60s) </li>
<li> 5-minute walk: Undirected walk, side step, and stand (~300s) </li>
<li> Speedskater: Speedskater on the spot (~30s) </li>
<li> Jog: Jog straight and return (~30s) </li>
<li> Jumping jacks: Jumping jacks on the spot (~30s) </li>
<li> High knee: High knee jog straight and return (~30s) </li>
<li> TUG: Timed up and go (~30s) </li>
</ul>

<p>neura-sparse.zip contains the main data (cut and aligned). neura-sparse-raw.zip contains the full trial.</p>

<p>Known Issues:</p>
<ul>
<li>Subject 10's left and right foot sensors were accidentally swapped. All have been fixed except for the xsens BVH output. from the MVN software. Note that Xsens MVN 2019.0, in theory, has the capability to swap sensors in reconstruction. However, the software crashes whenever I attempt to do so. I could not use a newer Xsens MVN version due to license limitation.</li>

</ul> </li>
</ul>