James D. Ratcliffe

Norm-Optimal Iterative Learning Control Applied to Gantry Robots for Automation Applications

This paper is concerned with the practical implementation of the norm-optimal iterative learning control (NOILC) algorithm. Here, the complexity of this algorithm is first considered with respect to real-time control applications, and a new modified version, fast norm-optimal ILC (F-NOILC), is derived for this application, which potentially allows implementation with a sampling rate three times faster that the original algorithm. A performance index is used to assess the experimental results obtained from applying F-NOILC to an industrial gantry robot system and, in particular, the effects of varying the parameters in the cost function, which is at the heart of the norm-optimal approach

Iterative learning control applied to a gantry robot and conveyor system

Synchronization is routinely required to coordinate the actions of the various sub-systems involved in process applications. This is commonly achieved through direct mechanical coupling, involving gears, drive belts and cams. Apart from the additional cost incurred, these components are subject to wear, constrain the layout of the plant and may have limited accuracy. It is shown in this paper that a mechanical linkage between two sub-systems may be replaced by instead implementing a control scheme comprising an iterative learning controller together with a supervisory control loop. To illustrate the approach, two types of iterative learning controller are first implemented on a gantry robot test facility to confirm the high levels of tracking accuracy that may be achieved. The supervisory control loop is then added to synchronize the ‘pick and place’ action of the robot with a conveyor system moving at constant velocity. Experimental results are provided to confirm both the accurate tracking performance produced by the iterative learning controller, and the high level of synchronization achieved by the overall scheme.

A new robust iterative learning control algorithm for application on a gantry robot

In this paper a new robust steepest-descent algorithm for discrete-time iterative learning control is introduced for plant models with multiplicative uncertainty. A theoretical analysis of the algorithm shows that if a tuning parameter in the algorithm is selected to be sufficiently large, the algorithm will result in monotonic convergence if the plant uncertainty satisfies a positivity condition. This is a major improvement when compared to the standard steepest-descent algorithm, which lacks a mechanism for finding a balance between convergence speed and robustness. Experimental work on a gantry robot is performed to demonstrate that the algorithm results in near perfect tracking in the limit.

Fast norm-optimal iterative learning control for industrial applications

Norm-optimal iterative learning control has potential to significantly increase the accuracy of many trajectory tracking tasks which can be found in industry. The algorithm can achieve very low levels of tracking error and the number of iterations required to reach minimal error is small compared to many other iterative learning control algorithms. However, in the current format, the algorithm is not attractive to industry because it requires a large number of calculations to be performed at each sample instant. This implies that control hardware must be very fast which is expensive, or that the sample frequency must be reduced which can result in reduced performance. To remedy these problems, a revised version, fast norm-optimal iterative learning control is proposed which is significantly simpler and faster to implement. The new version is tested both in simulation and in practice on a three axis industrial gantry robot.

Measuring the Performance of Iterative Learning Control Systems

Iterative learning control has the potential to significantly improve the tracking performance of repeating trajectory control systems. However, until now little attempt has been made to measure this performance quantitatively. A new iterative learning control performance index PIN is introduced which allows direct, quantitative, performance comparison of different algorithms, or alternatively a single algorithm which has variable tuning parameters. In particular, PI N can be used as a tool for selecting and adjusting algorithm tuning parameters. Application of the new performance index is demonstrated with both simulation studies and practical implementation on a gantry robot

Comparing the performance of two iterative Learning Controllers with optimal feedback control

The tracking performance of two iterative learning control algorithms is compared to that, which can be achieved by an optimal feedback controller. P-type iterative learning control in parallel with a proportional feedback controller is compared with norm-optimal iterative learning control, then both ILC systems are compared with the performance achieved by an optimal feedback controller. Considering that the ILC plus proportional controller requires no prior modelling of the plant and minimal adjustment of gains, the tracking performance in terms of mean squared error (mse) per iteration can be reduced by two orders of magnitude further than can be achieved with the optimal feedback controller. However, the norm-optimal ILC improves upon this performance by reducing the mse by an extra order of magnitude. The experimental results are derived from tests performed on a gantry robot

Iterative learning control - what is it all about?

Abstract The main objective of this paper is to show how one can benefit from using Iterative Learning Control instead of conventional feedback control. As a main result it is shown that even if the nominal plant satisfies a given uncertainty condition, there always exists ILC algorithms that can drive the tracking error monotonically to zero. This same result cannot be achieved with conventional feedback control, or by inverting a nominal model of the plant. Hence ILC offers an unique tool to invert dynamical systems with uncertainty.

Discrete-time Arimoto ILC-algorithm revisited

Abstract This paper revisits the Arimoto-algorithm in the discrete-time case. It is shown that if a plant satisfies a positivity condition, there always exists a learning gain so that the algorithm converges monotonically to zero tracking error. If the plant does not satisfy the positivity condition, a linear LQ tracker can be used to condition the plant so that it satisfies the positivity condition. The overall structure results in a novel combination of Arimoto ILC and LQ optimal control, that drives the tracking error monotonically to zero for an arbitrary discrete-time LTI plant. This is a very strong property for any ILC algorithm.

Practical implementation of a model inverse optimal iterative learning controller

Abstract In this paper, a new model inverse optimal iterative learning control algorithm is practically implemented on an industrial gantry robot. The algorithm has only one tuning parameter which can be adjusted to provide a balance between convergence speed and robustness. Results show that the algorithm is capable of learning the required trajectory in very few iterations. However at this convergence rate the lack of robustness is a major issue. Appropriate use of the tuning parameter is shown to greatly increase the algorithm robustness as demonstrated by tests which successfully reach 600 iterations.