Alberto F. Hamilton

A self-tuning neuromorphic controller: application to the crane problem

Abstract This paper is concerned with the design and application of a self-tuning controller, aided by means of neural network s (NN). The structure of the controller is based on the use of neural networks as an implicit self-tuner for the controller. The aim o f this approach is to take advantage of the learning properties of the neural networks to increase the performance of the self-tuning. The a pplication of this technique is performed on an overhead crane. The control objective is to suppress undesirable oscillations during op eration of the crane. First, some simulations were carried out, as well as a comparison with a standard self-tuning method, that demon strate the advantages of this method. After this, a real-time implementation on a scale prototype of a crane was done to verify th e applicability of the method.

A robotic system based on neural network controllers

Abstract In this paper, a control algorithm based on neural networks is presented. This control algorithm has been applied to a robot arm which has a highly nonlinear structure. The model based approaches for robot control (such as the computed torque technique) require high computational time and can result in a poor control performance, if the specific model-structure selected does not properly reflect all the dynamics. The control technique proposed here has provided satisfactory results. A decentralised model has been assumed here where a controller is associated with each joint and a separate neural network is used to adjust the parameters of each controller. Neural networks have been used to adjust the parameters of the controllers, being the outputs of the neural networks, the control parameters.

Design of a neural network based self-tuning controller for an overhead crane

In the process industry, the use of overhead crane systems for the transportation of material is very common. These are nonlinear systems that present undesirable oscillations during the motion, especially at arrival. The paper presents a self-tuning controller based on neural networks for the anti-swing control problem of the crane. The scheme of the controller is based on using neural networks as self-tuners for the parameters of a state feedback controller. The aim of this approach is to take advantage of the ability to learn of the neural networks and to use them in place of an identifier in the conventional self-tuner scheme. One of the main advantages of this method is that the training of the networks is done online using a backpropagation algorithm. The algorithm was implemented and tested by means of different simulations carried out with the crane.

Decoupling of the 10 m GranTeCan telescope's primary mirror dynamics and design of a controller for noise rejection

The goal of the paper is to present a method for the decoupling of a multivariable system and to show how this procedure allows the problem of vibrations rejection to be faced. The system we have worked with represents the dynamic behaviour of the GTCs (Gran Telescopio Canarias) 10 m diameter primary mirror. This telescope will be built in the island of La Palma (Spain). The big size of this mirror has caused its segmentation into 36 hexagonal pieces called segments. The system has 108 inputs and the same number of outputs coupled among them. With the decoupling method we present in this work, the problem of designing a controller for a MIMO system has been converted into the calculus of 108 SISO controllers. One of these controllers is presented as an example of how the decoupling process facilitates the noise rejection.

MCL with sensor fusion based on a weighting mechanism versus a particle generation approach

The combined action of several sensing systems, so that they are able to compensate the technical flaws of each other, is common in robotics. Monte Carlo Localization (MCL) is a popular technique used to estimate the pose of a mobile robot, which allows the fusion of heterogeneous sensor data. Several sensor fusion schemes have been proposed which include sensors like GPS to improve the performance of this algorithm. In this paper, an Adaptive MCL algorithm is used to combine data from wheel odometry, an inertial measurement unit, a global positioning system and laser scanning. A particle weighting model which integrates GPS measurements is proposed, and its performance is compared with a particle generation approach. Experiments were conducted on a real robotic car within an urban environment.

Stochastic optimal controllers for a DC servo motor: Applicability and performance

Abstract In this work some optimal control algorithms have been designed for and implemented in a real plant. The plant is a DC motor, controlled in the armature. Both deterministic and stochastic control policies have been developed. The aim of this paper is to show the applicability of optimal control algorithms to improve performances, to look for an empirical proof of the effects that appear in a stochastic policy (such as the caution effect), and to confirm an increment in the performance in the stochastic algorithms as compared with entirely deterministic algorithms. The control schemes applied are Dynamic Programming and Generalized Predictive Control.

Using neural networks to improve classification: application to brain maturation

The knowledge acquisition problem is one of the most difficult issues in elaborating a medical expert system. This is more true in the context of automated brain signal diagnosis. This kind of knowledge does not lend itself to be represented in a classical rule-based system and is not easily put in quantitative terms by the specialists. Artificial neural networks (ANNs) provide a useful alternative for capturing this information. In this work, an application of ANNs to brain maturation prediction is presented. The problem is essentially a supervised classification. A case data base consisting of data extracted from electroencephalographic (EEG) signals and diagnoses carried out by an expert neurologist serves to test the ability of several statistical classifiers and several kinds of ANNs in reproducing the expert results. There is also a discussion on how to integrate ANNs in a higher-level knowledge-based system for brain signal interpretation.

Towards a Multiagent Approach for the VERDINO Prototype

This paper presents a work in progress about the design and development of a multiagent system for an autonomous vehicle (VERDINO). This vehicle (a standard golf cart) has been provided with many different sensors and actuators. The future multiagent system is intended to manage the data provided by the sensors and act on steering orientation and brake and throttle pedals.

Experiments on A D.C. Motor Based System for a Digital Control Course

Experiments on a d.c. motor-based system for a digital control course A set of real-time experiments is presented. These experiments are implemented on a position-velocity system and cover a wide range of control techniques. We start with basic control actions and continue with more complex strategies like optimal control. We emphasize the study of all practical aspects of the experiences: noise, delays, etc.

MASCONTROL: a MAS for system identification and process control

In this paper, a MAS, called MASCONTROL for system identification and process control is presented, including an Ontology Agent. It implements a self- tuning regulator (STR) scheme. Defined classes, properties, axioms and individuals in the ontology are mainly related to control concepts. These definitions and other ones allow the system to make some interesting inferences from some axioms defined in the ontology.