ReadMe.txt for 'Adaptive Robot Mediated Upper Limb Training Using Electromyogram Based Muscle Fatigue Indicators'.

This dataset supports the following publication:
Azeemsha Thacham Poyil, Volker Steuber, Farshid Amirabdollahian. (2019) 'Adaptive Robot Mediated Upper Limb Training Using Electromyogram Based Muscle Fatigue Indicators', PLoS ONE.

Principal Investigator:
Azeemsha Thacham Poyil, School of Computer Science, University of Hertfordshire, Hatfield, AL10 9AB, United Kingdom; a.thacham-poyil@herts.ac.uk.

Co-Investigator and contact for queries relating to these data:
Prof. Farshid Amirabdollahian, School of Computer Science, University of Hertfordshire, Hatfield, AL10 9AB, United Kingdom; f.amirabdollahian2@herts.ac.uk.

This study formed part of the PhD program pursued by Azeemsha Thacham Poyil at University of Hertfordshire. 
The experiments were carried out in the adaptive systems lab in the School of Computer Science between 10/04/2018 and 30/5/2018. 

License:
We release the following documents under a creative commons 'CC-BY 4.0' license (https://creativecommons.org/licenses/by/4.0/):
* Readme.txt (this document)
* Electromyogram (EMG) data files named as "EMG_SubjectX.mat".
* Kinematic (HapticMaster robot) data files named as "Log_SubjectX.txt"


Description of Data:
The data set consists of kinematic and electromyogram (EMG) signals collected from 3 upper limb muscles (Biceps Brachii, Anterior Deltoid, and Middle Deltoid) of healthy participants during an adaptive robotic interaction involving rowing movements.
30 (17 males, 13 females) healthy participants of at least 18 years old with no history of injury to the upper limb and back were involved in this experiment.
Participants were students or staff members of University of Hertfordshire or other volunteers from outside the university.
The measurements were taken using an EMG acquisition device (g.USBamp amplifier) from g.tec medical engineering GmbH. 
The data acquisition parameters (sampling rate, channel selection and so on) of the device were configured using Simulink.
Three EMG electrode channels were configured in bipolar mode with a sampling frequency of 1200Hz. 
EMG electrodes for Biceps Brachii muscles were connected to Ports 11-12 of g.USBamp, Frontal Deltoid muscles to Ports 13-14 and Middle Deltoid muscles to Ports 15-16.
The tasks involved upper limb exercise, which simulated rowing, using a robot arm (HapticMaster robot) as directed by visual instructions on screen and audio cues. 
This involved moving a robotic end-effector, while an animated boat rowing environment running in front of the participant on an LCD monitor. 

There were be 3 groups of participants in this study. 
Control 1 Group (Group A) participants did not receive any adaptation from the robot during the interaction and were given break periods at regular intervals. 
Intervention Group (Group B) interacted with the adaptive robotic environment, which was designed to adjust the difficulty level of the training exercise based on EMG based fatigue indicators. 
Control 2 Group (Group C) participants had a similar environment as Group B, but the environment only adapted based on the subject-reported fatigue. 
The participants were asked to continue the exercise until they felt very tired or until they reported fatigue 3 times or until the maximum feasible robotic resistance was reached.

Those files with a '.mat' suffix are MATLAB output data files. Those files with a '.txt' suffix are log files containing the kinematic data from the robotic measurements.
For the purposes of archiving, variable names have been made consistent across datasets and the English language has been used.