Comprehensive Kinetic and EMG Dataset of Daily Locomotion with 6 types of Sensors

The paper is now published in the recent Wearable Technology Journal, more detailed information on this dataset can be found there!

A human movement experiment with 12 young adults performing 13 daily movement trials (6 walking trials (speed: 0.9, 1.8, 2.7, 3.6, 4.5, 5.4 km/  h; 3 running trials (speed: 6.3, 8.8, 9.9 km/h); and  four non-locomotion trials (vertical jump, squat, lunge, and single leg landing) was conducted with the ethic approved by the University of Twente (ET/A.21.19298, reference number 2021.57). 

Six types of measurement devices were used to capture different information of participants’ movements. They can be divided into two systems: the wearable system and the conventional non-wearable system. In the wearable measurement system, eight IMUs (Xsens Link, Enschede, The Netherlands) were used to measure the kinematic movements of lower limbs and trunk. A pair of pressure insoles (Moticon, Munich, Germany) was used to measure the vertical GRF and CoPs. In the conventional system, an optical motion capture system (OMC) containing 8 infrared light cameras (6+ series, Qualisys, Gothenburg, Sweden) was used to measure body kinematics data using reflective markers. A split-belt instrumented treadmill (Motek-Forcelink B.V, Culemborg, The Netherlands) was used to measure the GRFs under each foot. Two video cameras were also included inside the conventional system to capture the RGB images of participants’ body postures at the sagittal and frontal planes (https://doi.org/10.5281/zenodo.6644593). In addition, nine electromyography sensors (EMGs) (Delsys Trigno, Delsys, USA) were included to record the activations of nine major muscles in the dominant leg ("soleus", "medial gastrocnemius", "lateral gastrocnemius", "tibialis anterior", " semimembranosus", " biceps femoris long head", "vastus lateral", "rectus femoris", "vastus medial").

In this shared data repository, both raw data (to be uploaded) and processed data (Processed_data.zip) are provided. The data processing pipeline can be found in this public GitHub repository: https://github.com/ET-BE/BioMechPro/tree/study/WearableSystemValidation. Guidelines for creating the same wearable system in the corresponding comparison study are shared in this GitHub repo: https://github.com/HuaweiWang/WearableMeasurementSystem.

[Note!] If you have unstable network that not able to download the huge data files, please check this version of dataset with small file sizes(2GB each)

Raw data:  https://doi.org/10.5281/zenodo.7422043

Processed data:  https://doi.org/10.5281/zenodo.7422031 

Dataset structure descriptions:

Raw_data.rar: Raw dataset

Subjxx: subject folder

Qualisys: Qualisys project folder of the recordings

Xsens: Xsens project folder of the recordings

Insoles: Pressure insole project folder of the recordings

Processed_data.rar: Processed dataset.

allAverage.mat:  the overall summarization data of all subject at all movement trials. 

dataValidation.m: the Matlab code to plot the summarization data by loading the allAverage.mat.

subjs_info.txt: general information of all participants. 

ComparisonPlots: folder that contains the comparison plots between the laboratory-based system and the wearable measurement systems.

Subjxx: processed data for participants xx

dynMVCvalue.mat: the dynamic Maximal Voluntary Contraction of measured muscles (highest value among all recorded movements)

MVCvalue.mat: the Maximal Voluntary Contraction of measured muscles (highest value in MVC recording trial only)

Subjxx_xxxx_xx.mat: the processed data (generated by the above mentioned processing pipeline) of current subject at specific movement trial.

Qualisys: The exported mat files from the Qualisys recordings, including markers, EMGs, GRFs.

Xsens: The exported .mvnx files from the Xsens MVN software reprocessing. This file can be directly loaded by Matlab without requiring the Xsens license.

Insole: Insole recorded data, parsed from the Moticon endpoint SDK output.

OS: The folder that contains the scaled OpenSim model and corresponding .xml and data files for IK and ID processing. Majority content in this folder is automatically generated by the processing pipeline

Figures: plots of the processed data, including joint angles, GRFs, joint torques, and EMGs. They are all generated in the last module of the processing pipeline.

Structure of the Subjxx_xxxx_xx.mat:

Subjxx_xxxx_xx.mat:

Info: the general information of the participant and corresponding processing steps.

Marker: Marker data from Qualisys

Force: GRFs data from Qualisys

EMG: EMG data from Qualisys

IMU: Motion data from Xsens IMU system (from .mvnx)

Insole: The recorded pressure insole data

Resample: This data structure that contains the resampled data of the above mentioned sensor data

FrameRate: the sampling rate for all resampled sensor data

Marker: Resampled marker data

Force: resampled force data

EMG: resampled EMG data

IMU: resampled IMU data

Insole: resampled Insole data

CoM: resampled center of mass data from Xsens IMU system

Sych: this data structure contains the IK & ID data of two measurement systems. They are also synchronized by calculate the highest correlation coefficient. 

DeltaT: the time differences between the laboratory-based  system and the wearable measurement system.

IKAngData: the joint angle data from marker data inverse kinematics

ForcePlateGFRData: the ground reaction force data from instrumented treadmill

IDTrqData: the joint torque data from laboratory measurement system (optical + treadmill)

IMUAngData: the joint angle data from Xsens MVN software

InsoleGRFData: the ground reaction force data from pressure insoles

IDTrqData_portable: the joint torque data from the wearable system inverse dynamics

EMG: synchronized EMG data

CoM: synchronized CoM data

xxxxxLabel: the labels of each data column of corresponding data matrix

Average: this data structure contains the averaged gait/moment cycles

hsMatrix_right: The heel strike data points of the right leg

hsMatrix_left: the heel strike data points of the left leg

EMGAvedynNorFlag: whether dynamic MVC normalization is applied on EMG.

EMGAveNorFlag: whether MVC normalization is applied on EMG.

xxxx: The averaged data of corresponding variables

ForcePlateGRFDataInCalCn: transferred treadmill GRF data from the treadmill global coordinate to the local Calcaneus coordinate of the scaled OpenSim model.