Folder with contents that relate to system identification procedures applied to bicycling.
Peter's first report on system ID techniques for the bicycle model.
A preliminary attempt by Dr. Hess at time domain system ID for a bicyclist tracking a line. A pulsive step disturbance is applied for the roll torque. This uses Ron's manual control model.
Preliminary system ID with sensor input noise.
A Preliminary Study of a Rider/Bicycle Identification Technique Using a Computer Model of the Rider and Bicycle
Uses the Whipple model with a lateral force.
This is a first shot at identifying transfer functions from real data.
This is an ID attempt on a run of 60 seconds (#105) with about 7 pulls.
This is a transfer function identification for pull force to roll rate for a run (#105). The entire span of data (all 7 pulls) was used for the fit. The transfer function is then compared to our manual control model. The fit is remarkably good.
This is Ron's hypothesis on the human remnant that was found in some of the identification attempts.
This documents my attempts to setup various system identification models for the bicycle/rider system.
Various plots relating to the identification of the bicycle model.
I make use of the prediction error method (pem) in Matlab for system identification but don't really understand what it is doing. These are some tests to try to understand the options better.
These are Bode plots from the results of identifying the controller on 262 runs.
Bode plots of the 262 controller identifications showing all the lines.