Improvements in motion control are a key step towards meeting tightening requirements on throughput and accuracy in future industrial machines. Typically, a combination of traditional feedback and feedforward controllers is employed, which remain fixed throughout the life-cycle of the system. Learning control enables a significant performance increase over feedback and feedforward control by learning from past data, to anticipate and compensate for repeating components of the error. These disturbances include known setpoint signals and unknown disturbances, such as friction and hysteresis. Although important progress has been made in theory and design of learning controllers, its full potential for industrial mechatronic applications is largely unexploited [1]. This is partly due to several former shortcomings of learning control. The aim of this work is to provide a concise overview of recent advances in learning control that are particularly tailored towards successful industrial implementation, and reveal to the interested reader what learning control has to offer for his/her particular problem at hand.
[1]
Lennart Blanken,et al.
Flexible ILC : towards a convex approach for non-causal rational basis functions
,
2017
.
[2]
Lennart Blanken,et al.
Frequency domain design of iterative learning control and repetitive control for complex motion systems
,
2018
.
[3]
Andrew J. Fleming,et al.
Iterative control for periodic tasks with robustness considerations, applied to a nanopositioning stage
,
2016
.
[4]
Lennart Blanken,et al.
Multivariable repetitive control design framework applied to flatbed printing with continuous media flow
,
2017,
2017 IEEE 56th Annual Conference on Decision and Control (CDC).
[5]
Robin de Rozario,et al.
Improving transient learning behavior in model-free inversion-based iterative control with application to a desktop printer
,
2018,
2018 IEEE 15th International Workshop on Advanced Motion Control (AMC).