Optimization-based iterative learning control for robotic manipulators

Iterative learning control (ILC) has been intensely researched for over 30 years to improve the performance of repetitive processes. Most ILC algorithms use a known, but potentially inaccurate model to compute the next iteration’s control signal. The majority of publications on the topic of ILC considers linear-time-invariant or linear-parameter-varying systems, although many applications require nonlinear models to represent the system’s dynamics sufficiently. An example for such an application is a robotic manipulator executing the same task repeatedly. This paper adapts a general optimization-based ILC approach for arbitrary nonlinear systems to be used for manipulators with n degrees-of-freedom in a closed-loop configuration. The developed approach is validated both in simulation and experimentally for a 6 degrees-of-freedom robotic manipulator.status: publishe