gILC

gILC is open source software for a generic approach to iterative learning control (ILC) for nonlinear systems. Iterative learning control is a strategy for the open loop control of dynamic systems that need to perform a given task repeatedly. Its aim is to reject repeating disturbances and improve tracking control by using information about the tracking performance of the previous trial. gILC allows the user to tune the algorithm in a wide range of settings with minimal coding effort. It has excellent computational efficiency, even for long control tasks, and therefore reduces the required calculation time between trials.

gILC was originally developed by Marnix Volckaert in 2012. It is currently maintained and adapted for new research topics by Armin Steinhauser within the MECO research team. In the course of this adaptation, a Matlab translation of gILC was included.

Notes

Make sure to get CasADi (at least v3.0) before using gILC.

The originally published software (v1.3) can be found here. In case this website is not accessible its content is packed in the doc folder.

Version history

• v1.4, 09/2016

  • Changed to CasADi v3.0 syntax
  • Expecting separate CasADi installation and abandoning the install-paradigm
  • Include Matlab version
  • Move to Github repo

• v1.3, 05/2012

  • Updated manual document

• v1.2

  • Included option to provide the number of additional states (which won't be set equal to the initial state value)
  • Changed the default weight on tracking error to support multiple outputs
  • Fixed a bug that misinterpreted multiple references (for MIMO systems)
  • Removed additional constraints for ending the motion at rest using the relative degree
  • Changed the default weight on tracking error to 1e8
  • Updated tutorials

• v1.1

  • Updated tutorials 1,2 and 3

• v1.0

  • Initial version