search

csu-hmc / gait-control-direct-id-paper

Direct identification of the human's control scheme during walking.
http://hmc.csuohio.edu/projects/gait-control-id
4 stars 2 forks source link

Make sure random error in measurements doesn't give same results as the truly pertrubed motion. #6

Closed moorepants closed 10 years ago

moorepants commented 10 years ago

Does the data processing introduce a bias? The joint moment calculation makes use of marker coordinates. The joint angle calculation uses the same marker coordinates. If there was a random measurement error in a marker, it would affect both and this may result in correlations that show up in the sys id results. We should be able to test this effect and conclude whether it is a problem. Here is an idea: (1) make an average (periodic and unperturbed) gait cycle from raw data (marker coordinates and GRF) and repeat it over and over to create the same amount of data that you usually work with. (2) add small random numbers to all samples, (3) compute angles and torques in the same way you normally do, (4) do the sys id and see what comes out. If this result is the same as what you get from the perturbation experiment, that would be bad news...

moorepants commented 10 years ago

Done: f25ea7c97911cf560137fe307cbefe83582e8636