Evolution of neural control structures: some experiments on mobile robots

From perception to action and from action to perception, all elements of an autonomous agent are interdependent and need to be strongly coherent. The final behavior of the agent is the result of the global activity of this loop and every weakness or incoherence of a single element has strong consequences on the performances of the agent. We think that, for the purpose of building autonomous robots, all these elements need to be developed together in continuous interaction with the environment. We describe the implementation of a possible solution (artificial neural networks and genetic algorithms) on a real mobile robot through a set of three different experiments. We focus our attention on three different aspects of the control structure: perception, internal representation and action. In all the experiments these aspects are not considered as single processing elements, but as part of an agent. For every experiment, the advantages and disadvantages of this approach are presented and discussed. The results show that the combination of genetic algorithms and neural networks is a very interesting technique for the development of control structures in autonomous agents. The time necessary for evolution, on the other hand, is a very important limitation of the evolutionary approach.     

Evolutionary Optimisation of Mechanical Structures: Towards an Integrated Optimisation

This work results from the synthesis of author’s works on the applications of stochastic techniques (genetic algorithms with neural networks) for the optimisation of mechanical structures. The emphasis of this work is on the practical aspects and the feasibility of the aformentioned techniques. The research strategy consists in substituting, for finite element calculations in the optimisation process, an approximate response of a neural network. More precisely, the paper describes the use of backpropagation neural networks in creating function approximations for use in computationally intensive design optimisation based on genetic algorithms. An example of application for space frame optimisation of a helicopter tail boom is given in this paper, for which we can talk of integrated optimisation. This example (including displacement and frequency constraints) show the use of neural networks as a function approximation strategy to limit the computational costs associated with stochastic search methods.     

A neural network based multi-classifier system for gene identification in DNA sequences

The paper presents a neural network based multi-classifier system for the identification of Escherichia coli promoter sequences in strings of DNA. As each gene in DNA is preceded by a promoter sequence, the successful location of an E. coli promoter leads to the identification of the corresponding E. coli gene in the DNA sequence. A set of 324 known E. coli promoters and a set of 429 known non-promoter sequences were encoded using four different encoding methods. The encoded sequences were then used to train four different neural networks. The classification results of the four individual neural networks were then combined through an aggregation function, which used a variation of the logarithmic opinion pool method. The weights of this function were determined by a genetic algorithm. The multi-classifier system was then tested on 159 known promoter sequences and 171 non-promoter sequences not contained in the training set. The results obtained through this study proved that the same data set, when presented to neural networks in different forms, can provide slightly varying results. It also proves that when different opinions of more classifiers on the same input data are integrated within a multi-classifier system, we can obtain results that are better than the individual performances of the neural networks. The performances of our multi-classifier system outperform the results of other prediction systems for E. coli promoters developed so far.

Distributed multilevel optimization for complex structures

Optimization problems concerning complex structures with many design variables may entail an unacceptable computational cost. This problem can be reduced considerably with a multilevel approach: A structure consisting of several components is optimized as a whole (global) as well as on the component level. In this paper, an optimization method is discussed with applications in the assessment of the impact of new design considerations in the development of a structure. A strategy based on fully stressed design is applied for optimization problems in linear statics. A global model is used to calculate the interactions (e.g., loads) for each of the components. These components are then optimized using the prescribed interactions, followed by a new global calculation to update the interactions. Mixed discrete and continuous design variables as well as different design configurations are possible. An application of this strategy is presented in the form of the full optimization of a vertical tail plane center box of a generic large passenger aircraft. In linear dynamics, the parametrization of the component interactions is problematic due to the frequency dependence. Hence, a modified method is presented in which the speed of component mode synthesis is used to avoid this parametrization. This method is applied to a simple test case that originates from noise control.     

An artificial life approach for the animation of cognitive characters

This paper addresses the problem of cognitive character animation. We propose the use of finite state machines for the behavioral control of characters. Our approach rests on the idea that the cognitive character arises from the evolutionary computation embedded in the artificial life simulation, which in our case is implemented by the finite state machine. We present some of the results of the WOXBOT/ARENA research project. This project to build virtual worlds is aimed at the graphic simulation of an arena, where small mobile robots can perform requested tasks while behaving according to their own motivation and reasoning. Each robot is an intelligent agent that perceives the virtual environment through a simulated vision system and reacts by moving away from or approaching the object it sees. The conception and specification of the robots and environment are being done very carefully to create an open distributed object architecture that could serve as a test-bed freely available and ready to use for testing theories in some computational areas such as evolutionary computation, artificial life, pattern recognition, artificial intelligence, cognitive neurosciences and distributed objects architectures. Furthermore, it is a first step towards building a cognitive animated character.     

基于改进型遗传算法的前馈神经网络优化设计

阐明了遗传算法和神经网络结合的可行性，提出了一种改进的面向神经网络权值学习的遗传算法。通过对XOR问题的实验，显示出其快速学习网络权值的能力，且能摆脱局部极值的困扰和初始权值的限制，从各方面都表现出优于标准遗传算法和BP算法的性能。     

SID-GA: An evolutionary approach for improving observability and redundancy analysis in structural instrumentation design

In this paper the core of a genetic algorithm designed to define a sensor network for instrumentation design (ID) is presented. The tool has been incorporated into a decision support system (DSS) that assists the engineer during the ID process. The algorithm satisfactorily deals with non-linear mathematical models, and considers four design objectives, namely observability, cost, reliability and redundancy, exhibiting properties that were either never addressed by existing techniques or partially dealt with in the literature. Its performance was tested by carrying out the ID of an ammonia synthesis industrial plant. Results were statistically analysed. A face validity study on the fitness function’s soundness was also assessed by a chemical engineer with insight and expertise in this problem. The technique performed satisfactorily from the point of view of the expert in ID, and therefore it constitutes a significant upgrading for the DSS.     

Evolutionary Multiobjective Design in Automotive Development

This paper describes the use of evolutionary algorithms to solve multiobjective optimization problems arising at different stages in the automotive design process. The problems considered are black box optimization scenarios: definitions of the decision space and the design objectives are given, together with a procedure to evaluate any decision alternative with regard to the design objectives, e.g., a simulation model. However, no further information about the objective function is available. In order to provide a practical introduction to the use of multiobjective evolutionary algorithms, this article explores the three following case studies: design space exploration of road trains, parameter optimization of adaptive cruise controllers, and multiobjective system identification. In addition, selected research topics in evolutionary multiobjective optimization will be illustrated along with each case study, highlighting the practical relevance of the theoretical results through real-world application examples. The algorithms used in these studies were implemented based on the PISA (Platform and Programming Language Independent Interface for Search Algorithm) framework. Besides helping to structure the presentation of different algorithms in a coherent way, PISA also reduces the implementation effort considerably.     

Structure optimization of neural networks for evolutionary design optimization

We study the use of neural networks as approximate models for the fitness evaluation in evolutionary design optimization. To improve the quality of the neural network models, structure optimization of these networks is performed with respect to two different criteria: One is the commonly used approximation error with respect to all available data, and the other is the ability of the networks to learn different problems of a common class of problems fast and with high accuracy. Simulation results from turbine blade optimizations using the structurally optimized neural network models are presented to show that the performance of the models can be improved significantly through structure optimization.We would like to thank the BMBF, grant LOKI, number 01 IB 001 C, for their financial support of our research.     

基于生态竞争模型的遗传强化学习

未成熟收敛和收敛速度慢是目前遗传算法的明显缺点.借鉴生物在环境生态系统中的生长模式,文章提出一种生态竞争模型.该模型认为,竞争行为在生物的成长中占有十分重要的地位,在子群内实现了个体层次的先天遗传进化和后天竞争学习,在种群层次实现进一步的竞争强化学习.实验结果显示了该模型在解决收敛性问题时的有效性.     

用遗传BP网络进行图像边缘检测

该文提出了一种基于遗传算法与图像特征向量的边缘检测方法。由于噪声的干扰,常规的图像边缘检测方法往往效果不佳,因此在充分考虑边缘和噪声本质区别的基础上,构造具有较强抗噪能力的特征向量;然后用样本图像对多层前馈神经网络采用遗传学习算法和误差反向传播算法(BP)相结合进行训练,即先用遗传学习算法进行全局训练,再用BP算法进行精确训练,使网络收敛速度加快和避免局部极小。最后,将训练后的网络用于图像的边缘检测。实验证明这种方法是有效的。     

Development of a parsimonious GA-NN ensemble model with a case study for Charpy impact energy prediction

A parsimonious genetic algorithm guided neural network ensemble modelling strategy is presented. Each neural network candidate model to participate in the ensemble model is structurally selected using a genetic algorithm. This provides an effective route to improve the performance of the individual neural network models as compared to more traditional neural network modelling approaches, whereby the neural network structure is selected through some trial-and-error methods or heuristics. The parsimonious neural network ensemble modelling strategy developed in this paper is highly efficient and requires very little extra computation for developing the ensemble model, thus overcoming one of the major known obstacles for developing an ensemble model. The key techniques behind the implementation of the ensemble model, include the formulation of the fitness function, the generation of the qualified neural network candidate models, as well as the specific definitions of the assemble strategies. A case study is presented which exploits a complex industrial data set relating to the Charpy impact energy for heat-treated steels, which was provided by Tata Steel Europe. Modelling results show a significant performance improvement over the previously developed models for the same data set.     

An Incremental Distributed Algorithm for Computing Biconnected Components in Dynamic Graphs

This paper describes a distributed algorithm for computing the biconnected components of a dynamically changing graph. Our algorithm has a worst-case communication complexity of O(b+c) messages for an edge insertion and O(b'+c) messages for an edge removal, and a worst-case time complexity of O(c) for both operations, where c is the maximum number of biconnected components in any of the connected components during the operation, b is the number of nodes in the biconnected component containing the new edge, and b' is the number of nodes in the biconnected component just before the deletion. The algorithm is presented in two stages. First, a serial algorithm is presented in which topology updates occur one at a time. Then, building on the serial algorithm, an algorithm is presented in which concurrent update requests are serialized within each connected component. The problem is motivated by the need to implement causal ordering of messages efficiently in a dynamically changing communication structure. Received January 1995; revised February 1997.     

Design optimization of a spatial six degree-of-freedom parallel manipulator based on artificial intelligence approaches

Optimizing the system stiffness and dexterity of parallel manipulators by adjusting the geometrical parameters can be a difficult and time-consuming endeavor, especially when the variables are diverse and the objective functions are excessively complex. However, optimization techniques that are based on artificial intelligence approaches can be an effective solution for addressing this issue. Accordingly, this paper describes the implementation of genetic algorithms and artificial neural networks as an intelligent optimization tool for the dimensional synthesis of the spatial six degree-of-freedom (DOF) parallel manipulator. The objective functions of system stiffness and dexterity are derived according to kinematic analysis of the parallel mechanism. In particular, the neural network-based standard backpropagation learning algorithm and the Levenberg–Marquardt algorithm are utilized to approximate the analytical solutions of system stiffness and dexterity. Subsequently, genetic algorithms are derived from the objective functions described by the trained neural networks, which model various performance solutions. The multi-objective optimization (MOO) of performance indices is established by searching the Pareto-optimal frontier sets in the solution space. Consequently, the effectiveness of this method is validated by simulation.     

Design of an adaptive mutation operator in an electrical load management case study

An adequately designed and parameterized set of operators is crucial for an efficient behaviour of Genetic Algorithms (GAs). Several strategies have been adopted in order to better adapt parameters to the problem under resolution and to increase the algorithm's performance. One of these approaches consists in using operators presenting a dynamic behaviour, that is displaying a different qualitative behaviour in different stages of the evolutionary process. In this work a comparative analysis of the effects of using an adaptive mutation operator is presented in the operational framework of a multi-objective GA for the design and selection of electrical load management strategies. It is shown that the use of a time/space varying mutation operator depending on the values achieved for each objective function increases the performance of the algorithm.     

Application of neural network in suppressing mechanical vibration of a permanent magnet linear motor

In this study, a recurrent neural network compensator for suppressing mechanical vibration in a permanent magnet linear synchronous motor (PMLSM) is studied. The linear motor is controlled by a conventional PI velocity controller, and the vibration of the flexible mechanism is suppressed by using a hybrid recurrent neural network. The differential evolution strategy and Kalman filter method are used to avoid the local minimum problem, and estimate the states of system, respectively. The proposed control method is firstly designed by using a nonlinear simulation model built in Matlab Simulink and then implemented in a practical test rig. The proposed method works satisfactorily and suppresses the vibration successfully.     

Procedures for estimating desirable initial states of a production line: a comparative study

Two procedures for estimating initial states of a production line that ensure the line has a high probability of meeting the specified production target during a scheduled production shift are presented. The problem of determining desirable initial states is important in low variety, high volume production systems such as those from the automobile industry. One procedure is derived from design of experiments (DOE) theory whereas the other uses a genetic algorithm (GA). In the study it was determined that both procedures are straightforward to implement and produce good solutions to the problem. The results from the procedures are compared and their benefits and disadvantages are discussed.     

Integrated intelligent computing paradigm for the dynamics of micropolar fluid flow with heat transfer in a permeable walled channel

In this work, a novel application of bio-inspired computational heuristic paradigm is presented for micropolar fluid flow and heat transfer system in a channel with permeable walls by modeling competency of neural networks, global search of genetic algorithms, and rapid local convergence of sequential quadratic programming. Approximation theory in the mean squared error sense is exploited for the formulation of an objective function to solve the governing nonlinear fluidics system. The designed scheme is employed to study the dynamics of the model in terms of stream function, microrotation, concentration, and temperature profiles for prominent factors based on Reynolds number, Peclet number for diffusion of heat and mass, coupling, spin-gradient viscosity, micro inertia density parameters. The consistency and robustness of the solver are validated through statistical performance indices based on comparison with state of art Adams numerical method for accuracy and complexity measures.     

Feature selection by genetic algorithms in object-based classification of IKONOS imagery for forest mapping in Flanders, Belgium

Obtaining detailed information about the amount of forest cover is an important issue for governmental policy and forest management. This paper presents a new approach to update the Flemish Forest Map using IKONOS imagery. The proposed method is a three-step object-oriented classification routine that involves the integration of 1) image segmentation, 2) feature selection by Genetic Algorithms (GAs) and 3) joint Neural Network (NN) based object-classification. The added value of feature selection and neural network combination is investigated. Results show that, with GA-feature selection, the mean classification accuracy (in terms of Kappa Index of Agreement) is significantly higher (p < 0.01) than without feature selection. On average, the summed output of 50 networks provided a significantly higher (p < 0.01) classification accuracy than the mean output of 50 individual networks. Finally, the proposed classification routine yields a significantly higher (p < 0.01) classification accuracy as compared with a strategy without feature selection and joint network output. In addition, the proposed method showed its potential when few training data were available.     

A novel soft cluster neural network for the classification of suspicious areas in digital mammograms

This paper presents a novel soft cluster neural network technique for the classification of suspicious areas in digital mammograms. The technique introduces the concept of soft clusters within a neural network layer and combines them with least squares for optimising neural network weights. The idea of soft clusters is proposed in order to increase the generalisation ability of the neural network by providing a mechanism to more aptly depict the relationship between the input features and the subsequent classification as either a benign or malignant class. Soft clusters with least squares make the training process faster and avoid iterative processes which have many problems. The proposed neural network technique has been tested on the DDSM benchmark database. The results are analysed and discussed in this paper.