Walking Data README
Contact: Jared Markowitz   (jared.markowitz@gmail.com, 610-739-0107)

This directory contains walking data from three different collections:

1.)  The first set is a folder called "holodeck" and was taken with the force plate setup there in 2009.  Given the later, better options I would discourage its use.  

2.)  The second set is in a folder called "treadmill_no_metabolics".  This contains four sets collected at the Harvard Skeletal Biology Lab in 2011 and includes motion capture, force plate, and EMG data for subjects walking at 6-7 speeds (0.75m/s -> 2.0 m/s by 0.25 m/s, sometimes one extra "self-selected" speed).  The muscles for which EMG data were collected in this set were TA, SOL, GAS, VAS, BFSH, RF, HAM, ILL, GMAX.  Out of this set, AD052611 and AM062411 are the best sets.  DS100611 is fine but the subject's self-selected walking speed was a bit slow (~ 1.00 m/s).  EM072611 exhibits an interesting "bouncy" gait and would be good for testing robustness of models.

3.)  The third set is in a folder called "treadmill_full."  It is the best of the three and includes five sets collected in late 2012 / early 2013 at the Harvard Skeletal Biology Lab.  It includes motion capture, force plate, EMG, and metabolic data for subjects walking at  7 speeds (0.75m/s -> 2.0 m/s by 0.25 m/s as well as the speed where metabolic cost of transport was minimal).  The muscles for which EMG data were collected in this set were TA, SOL, GAS, VAS, BFSH, RF, HAM, ILL, GMAX, GMED, ADDL, ADDM.  In general the EMG data from the monoarticular muscles spanning the hip are lackluster.  All of these data sets are pretty good and you can see my thesis for further details.  One small omission: basal metabolic data were not collected for DH121412 or MC121812.  However the basal rates (normalized by body mass) seem to be closely clustered so using the average of the other three sets is likely fine for most applications.

-> The treadmill data set directories include a folder for each participant with both raw and processed data for the various collected modalities.  Most of the folders and files therein are labelled so that it's obvious what they are, but I'm happy to answer questions.  Each participant's data folder set also contains a "scaling" subfolder that includes the SIMM .jnt and .msl muscles used for processing.  The parent directory has a "metabolic_data" directory that includes the metabolic output for each participant as well as a SIMM directory that has additional useful files.

The other directories here are:

processing - contains all of my data processing scripts.  They won't be directly applicable because some of the paths are hard-coded but can be utilized with minor modifications.  In particular the script 'processing/processEMG/gc_avg/select_curves.m' is generally useful for removing outlier gait cycles from a data set.

miscellaneous- other files of miscellaneous use including some movies of SIMM in action made for Dr. Herr and digital (.mat) representations of the muscle stimulation data in Perry's Gait Analysis book (1992).