Patrícia Amâncio Vargas

Artificial homeostatic system: a novel approach

Many researchers are developing frameworks inspired by natural, especially biological, systems to solve complex real-world problems. This work extends previous work in the field of biologically inspired computing, proposing an artificial endocrine system for autonomous robot navigation. Having intrinsic self-organizing behaviour, the novel artificial endocrine system can be applied to a wide range of problems, particularly those that involve decision making under changing environmental conditions, such as autonomous robot navigation. This work draws on “embodied cognitive science”, including the study of intelligence, adaptivity, homeostasis, and the dynamic aspects of cognition, in order to help lay down fundamental principles and techniques for a novel approach to more biologically plausible artificial homeostatic systems. Results from using the artificial endocrine system to control a simulated robot are presented.

An immune learning classifier network for autonomous navigation

This paper proposes a non-parametric hybrid system for autonomous navigation combining the strengths of learning classifier systems, evolutionary algorithms, and an immune network model. The system proposed is basically an immune network of classifiers, named CLARINET. CLARINET has three degrees of freedom: the attributes that define the network cells (classifiers) are dynamically adjusted to a changing environment; the network connections are evolved using an evolutionary algorithm; and the concentration of network nodes is varied following a continuous dynamic model of an immune network. CLARINET is described in detail, and the resultant hybrid system demonstrated effectiveness and robustness in the experiments performed, involving the computational simulation of robotic autonomous navigation.

Preliminary investigations on the evolvability of a non-spatial GasNet model

This paper addresses the role of space in evolving a novel Non-Spatial GasNet model. It illustrates that this particular neural network model which make use of modulatory effects of diffusing gases has its evolvability improved when its neurons are not constrained to a Euclidean space. The results show that successful behaviour is achieved in fewer evaluations for the novel unconstrained GasNet than for the original model.

Mapping Artificial Immune Systems into Learning Classifier Systems

This paper presents one form of mapping Artificial Immune Systems (AIS) into Learning Classifier Systems (LCS). Artificial Immune Systems can be defined as adaptive systems inspired by theoretical models and principles of the biological immune system and applied to solve problems in the most diverse domains, from biology to computing. Similar to Learning Classifier Systems, already used to model complex adaptive systems, a better understanding of Artificial Immune Systems can be obtained when they are analysed under the perspective of complex adaptive systems. One of the goals here is to determine complementary features of both systems (LCS and AIS), aiming at providing a novel mapping conception. The formal treatment proposed along the paper may then be used to integrate models for complex adaptive systems.

On-Line Approach for Loss Reduction in Electric Power Distribution Networks Using Learning Classifier Systems

The problem of minimization of energy losses in power distribution systems can be formulated as obtaining the best network configuration, through the manipulation of sectionalizing switches. Using graph terminology, we have a combinatorial optimization problem, whose solution corresponds to finding a minimum spanning tree for the network. As an on-line approach to loss reduction in electric power distribution networks, this paper relies on Learning Classifier Systems to continually proposed network configurations close to the one associated with minimum energy losses, in the case of time-varying profiles of energy requirement. In order to evolve the set of rules that composes the Classifier System, operators for selection, reproduction and mutation are applied. Case studies illustrate the possibilities of this approach.

Evolving an Artificial Homeostatic System

Theory presented by Ashby states that the process of homeostasis is directly related to intelligence and to the ability of an individual in successfully adapting to dynamic environments or disruptions. This paper presents an artificial homeostatic system under evolutionary control, composed of an extended model of the GasNet artificial neural network framework, named NSGasNet, and an artificial endocrine system. Mimicking properties of the neuro-endocrine interaction, the system is shown to be able to properly coordinate the behaviour of a simulated agent that presents internal dynamics and is devoted to explore the scenario without endangering its essential organization. Moreover, sensorimotor disruptions are applied, impelling the system to adapt in order to maintain some variables within limits, ensuring the agent survival. It is envisaged that the proposed framework is a step towards the design of a generic model for coordinating more complex behaviours, and potentially coping with further severe disruptions.

Analysing Learning Classifier Systems in Reactive and Non-reactive Robotic Tasks

There are few contributions to robot autonomous navigation applying Learning Classifier Systems (LCS) to date. The primary objective of this work is to analyse the performance of the strength-based LCS and the accuracy-based LCS, named EXtended Learning Classifier System (XCS), when applied to two distinct robotic tasks. The first task is purely reactive, which means that the action to be performed can rely only on the current status of the sensors. The second one is non-reactive, which means that the robot might use some kind of memory to be able to deal with aliasing states. This work presents a rule evolution analysis, giving examples of evolved populations and their peculiarities for both systems. A review of LCS derivatives in robotics is provided together with a discussion of the main findings and an outline of future investigations.

Application of Learning Classifier Systems to the On-Line Reconfiguration of Electric Power Distribution Networks

This work explores the use of Learning Classifier Systems (Holland, 1992), to obtain good solutions to the problem of loss reduction in electric power distribution systems with time-varying demands. Loss reduction in power distribution systems can be achieved by changing the status of distribution switches towards alternative power flows. Given a demand profile, finding the best status for switches is a difficult combinatorial problem. To date, no approach is guaranteed to achieve an optimal solution for real scale distribution systems in a reasonable time-scale. However, heuristic approaches have been able to provide approximate solutions. Larger gains can be achieved if on-line reconfiguration is allowed in order to consider varying demand profiles (Lee and Brooks, 1988). Few papers have attempted to address this dynamic problem (Vargas et al., 2001; Zhou et al., 1997), and additional adaptation skills are necessary when compared with the corresponding static problem.

Redução de perdas em redes de distribuição de energia elétrica através de sistemas classificadores

The problem of minimising the technical energy losses in electric power distribution systems corresponds to the definition of the best network configuration, through the manipulation of sectionalising switches. Using a graph terminology, we have a combinatorial optimisation problem, whose solution asks for the definition of the minimum spanning tree for the network. With the purpose of treating the case of time-varying profiles of energy requirements, this paper presents the first results obtained with the application of learning classifier systems, an evolutionary computation approach to find network configurations close to the one associated with minimum energy losses. Its uttermost feature is the possibility of application to on-line network supervision, proposing new configurations that will reduce the level of energy losses when significant demand variations are detected.