Showing papers in "Artificial Life and Robotics in 1998"

Chaotic local search algorithm

Abstract: The steepest descent search algorithm is modified in conjunction withchaos to solve the optimization problem of an unstructured search space. The problem is that given only the gradient information of the quality function at the present configuration,X(t), we must find the value of a configuration vector that minimizes the quality function. The proposed algorithm starts basically from the steepest descent search technique but at the prescribed points, i.e., local minimum points, the chaotic jump is performed by the dynamics of a chaotic neuron. Chaotic motions are mainly caused because the Gaussian function has a hysteresis as a refractoriness. An adaptation mechanism to adjust the size of the chaotic jump is also given. In order to enhance the probability of finding the global minimum, a parallel search strategy is developed. The validity of the proposed method is verified in simulation examples of the function minimization problem and the motion planning problem of a mobile robot.

Dual dynamics: Designing behavior systems for autonomous robots

Abstract: This paper describes the “dual dynamics” (DD) design scheme for robotic behavior control systems. Behaviors are formally specified as dynamical systems using differential equations. A key idea for the DD scheme is that a robotic agent can work in different “modes,” which lead to qualitatively different behavioral patterns. Mathematically, transitions between modes are bifurcations in the control system.

Cooperative behavior of interacting robots

Abstract: We have researched the efficiency of cooperative behavior of interacting multirobots. In this paper, we generalize the definition of swarm intelligence and examine the emergence of the generalized swarm intelligence (here we call it “swarm function”) through the task of gathering pucks in a field by interacting simple robots. This robot has a drive system and the simplest means of interaction. The effectiveness of group behavior was studied for various (homogeneous, localized) puck distributions. To evaluate the efficiency of group behavior, we proposed a scaling relation between the task completion time and the number of robots, and examined the relation between the interaction duration and the efficiency of the group. We also proposed a simplified state transition diagram of the group and analysed their characteristics using it.

A control design of robotics using the genetic algorithm

Abstract: This paper presents a new and practical method for a control design of a robotic system. In general, actuators in robotic systems are set with gears whose characteristics are elastic. Since a state feedback-type digital controller is usually used for such a robotic system, the design of the feedback gain of the controller is important, because undesirable vibrations or an overshoot in responses occur for high gains. Therefore the desired response, the output of a reference model, is designed first, and the feedback gains are determined so that the response will coincide with the desired response, which is an optimization problem. The gradient method works to some extent, but it takes a long time to get a satisfactory result. Thus we applied the genetic algorithm (GA) to this nonlinear optimization problem, which gave the very first convergence. The gains obtained have many useful applications. The results of a simulation are also given.

Learning model for adaptive behaviors as an organized group of swarm robots

Abstract: This paper describes a novel organizational learning model for multiple adaptive robots. In this model, robots acquire their own appropriate functions through local interactions among their neighbors, and get out of deadlock situations without explicit control mechanisms or communication methods. Robots also complete given tasks by forming an organizational structure, and improve their organizational performance. We focus on the emergent processes of collective behaviors in multiple robots, and discuss how to control these behaviors with only local evaluation functions, rather than with a centralized control system. Intensive simulations of truss construction by multiple robots gave the following experimental results: (1) robots in our model acquire their own appropriate functions and get out of deadlock situations without explicit control mechanisms or communication methods; (2) robots form an organizational structure which completes given tasks in fewer steps than are needed with a centralized control mechanism.

Distributed and autonomous sensing based on the immune network

Abstract: A distributed and autonomous sensor network is proposed based on the informational features of the immune network: recognition of nonself by distributed and dynamically interacting units, recognition by a simple comparison with the units themselves, dynamic propagation of activation that would lead to system-level recognition, and memory embedded as stable equilibrium states in the dynamic network. The network is explained by an illustrative example of an eight-coin puzzle: a balance must be used only three times to identify one coin with a different weight from the other seven coins. Our network also uses a dynamic structure network rather than the fixed structures used in neural networks. Simulations show that nonself (the different coin in the eight-coin puzzle, the sensor/process fault in the monitoring example) will be identified by dynamically propagating activation through the network.

Development of a stair-climbing mobile robot with legs and wheels

Abstract: This paper deals with the development of a stair-climbing mobile robot with legs and wheels. The main technical issues in developing this type of robot are the stability and speed of the robot while climbing stairs. The robot has two wheels in the front of the body to support its weight when it moves on flat terrain, and it also has arms between the wheels to hook onto the tread of stairs. There are two pairs of legs in the rear of the body. Using not only the rorational torque of the arms and the wheels, but also the force of the legs, the robot goes up and down stairs. It measures the size of stairs when going up and down the first step, and therefore the measurement process does not cause this robot to lose any time. The computer which controls the motion of the robot needs no complicated calculations as other legged robots do. The mechanism of this robot and the control algorithm are described in this paper. This robot will be developed as a wheelchair with a stair climbing mechanism for disabled and elderly people in the near future.

“CAM-brain” ATR's billion neuron artificial brain project: A three-year progress report

Abstract: This paper reports on recent progress made in ATR's attempt to build a 10 000-module evolved neural net artificial brain to control the behavior of a life-sized robot kitten.

The application of artificial life to interactive computer installations

Abstract: This paper reports on the creation of interactive computer installations that combine artificial life and real life by means of human-computer interactions. These installations have focused on real-time interactions and evolutionary image processes. Accordingly, visitors to the installations become essential parts of the systems by transmitting their individual behaviors, emotions, and personalities to the image processes of the work. Images in these installations are no longer static, pre-fixed, and predictable, but become “living systems” themselves, representing minute changes in the viewers' interactions with the evolutionary image processes. Natural evolution has brought about a vast variety of forms and structures in nature. This research considers how artificial evoluation can function as a tool of the visual creation process; design should no longer be done by a designer or artist, but should emerge through the evolutionary image process itself.

A genetic algorithm and its parallelization for graph matching with similarity measures

Abstract: Graph matching and similarity measures of graphs have many applications to pattern recognition, machine vision in robotics, and similarity-based approximate reasoning in artificial intelligence. This paper proposes a method of matching and a similarity measure between two directed labeled graphs. We define the degree of similarity, the similar correspondence, and the similarity map which denotes the matching between the graphs. As an approximate computing method, we apply genetic algorithms (GA) to find a similarity map and compute the degree of similarity between graphs. For speed, we make parallel implementations in almost all steps of the GA. We have implemented the sequential GA and the parallel GA in C programs, and made simulations for both GAs. The simulation results show that our method is efficient and useful.

The improved radial basis function neural network and its application

Abstract: Compared with other feed-forward neural networks, radial basis function neural networks (RBFNN) have many advantages which make them more suitable for nonlinear system modeling, and they have recently received considerable attention. In this paper, a RBFNN is employed to model strongly nonlinear systems. First, the problems of nonlinear system modeling are analyzed, and then the structure of the RBFNN as well as the training algorithm are improved to solve these problems. Finally, an industrial high-purity distillation column, which is a strongly nonlinear system, is successfully modeled with the improved RBFNN. Owing to the complexities of a nonlinear system, it is necessary to use a real-time model correction method to modify the parameters of the RBFNN model in real time. One efficient method is proposed in this paper. The idea is to employ the Givens transformation to modify the parameters of the RBFNN-based model.

Progressive evolution model using a hardware evolution system

Abstract: A progressive evolution model is proposed in which evolution takes place stepwise to match environmental changes. This model was designed to accelerate evolution. Environmental complexity is defined, and the problem environment progresses in environmental complexity gradually from easy to difficult. A verification system for the model is constructed on a hardware evolution system called AdAM (adaptive architecture methodology) in which each individual circuit takes parallel input sequences and operates on this input. A measure that is suitable for such parallel simultaneous operations schemes is designed to express environmental complexity. Simulations using an artificial ant problem (a modified John Muir trail) show that in the progressive evolution model, circuits can easily evolve complex behaviors.

HOJO-brain for motion control of robots and biological systems

Abstract: The purpose of this research was to propose and develop a control method in the robotic and biomedical fields which is configured by a robotic/biological simulator, an analytical control frame which has phase sequences, sensory feedback, and an artificial central pattern generator (CPG) which is constructed by a recurrent neural network (RNN) and a genetic algorithm (GA). We call such a controller a “HOJO-brain”, which means a supplementary brain for motion control. We applied this method in the robotic and biomedical fields. In the robotic field, the HOJO-brain was applied to a 5-DOF legged-locomotion robot and a 32-DOF humanoid simulation model consisting of antagonistic muscles. In the biomedical field, it was applied to animals as the FES (functional electrical stimulation) controller. This FES control system with a HOJO-brain has the potential to give more effective and emergent motion control to severely physically handicapped people such as quadraplegics. With computer simulations and simple experiments using animals, we abtained performance indices which confirmed the fine adaptability and emergence for motion control.

Fine motion strategy in three-dimensional space using skill-based backprojection

Abstract: The motion of manipulation in a task can be decomposed into several motion primitives called “skills.” Skill-based motion planning gives the possibility of performing tasks as skillfully as human beings do. On the other hand, the backprojection method performed in configuration space has often been used in fine-motion planning. This paper describes fine-motion planning in three-dimensional space using skill-based backprojection. Now that skill-based planning in three-dimensional space has been developed, it becomes possible to plan manipulation motions like the behavior of the human hand.

A self-organizing genetic algorithm for multimodal function optimization

Abstract: A genetic algorithm (GA) has control parameters that must be determined before execution. We propose a self-organizing genetic algorithm (SOGA) as a multimodal function optimizer which sets GA parameters such as population size, crossover probability, and mutation probability adaptively during the execution of a genetic algorithm. In SOGA, GA parameters change according to the fitnesses of individuals. SOGA and other approaches for adapting operator probabilities in GAs are discussed. The validity of the proposed algorithm is verified in simulation examples, including system identification.

Cortical architecture and self-referential control for brain-like processing in artificial neural systems

Abstract: Progress in understanding the way the brain processes information while it is constantly interacting with the sensory environment is hampered by inadequate models and theories. Current models and theories of brain computing are, obviously, still not completely correct when confronted with so-called real-world problems. Sensory recognition and the subsequent selection and optimization of a proper behavior are basically constraint satisfaction problems. Both conventional AI and current formal neural network systems operate with set constraints: the architecture and parameters are defined a priori and then the input data are structured according to these set constraints on the learning process. However, as long as the constraints are set from outside the system (by the programmer, designer), the system has no ability for self-organization. There is the ability for adaptation within these a priori defined limits, but not the ability to include new knowledge into the consistent relational framework of existing knowledge beyond the prespecified constraints. Therefore, self-organization of constraints in complex systems is the key problem for getting self-organization of knowledge representation under real-world conditions. We show that a value system and self-referential control in a modular architecture are crucial prerequisites for both robust recognition of sensory input and the ability to integrate new knowledge into the already acquired knowledge representation. Finally, we outline a philosophy and propose a model approach that is a first step toward implementing those capabilities in artificial neural systems.

Intelligent control with new image processing strategy for a mobile vehicle

Abstract: This paper considers the problem of how to use a camera to recognize the shape of a path without using too much image processing time. A quadrangle image scene is divided into three parts, and the center of gravity of an object in each part is extracted to estimate the shape of the path. A strategy for how to measure distances with a camera is also presented. The idea behind this strategy is first to establish a methematical model describing the relationship between pixel distance in the image scene and real distance in front of the mobile vehicle by experiments, and then to decide the relevant position of the object by means of rotation of the camera. These image processing methods can be used in control problems with mobile vehicles/car-like robots, and can be used in fuzzy control, neural control, and other control strategies.

From sensory substitute technology to virtual reality research

Abstract: This paper describes research into sensory substitutes for the disabled, and proposes one basic research approach to assistance technology. It also reports on how the technology is related to virtual reality research.

Self-organization based on nonlinear nonequilibrium. Dynamics of autonomous agents

Abstract: This paper discusses a study on the mechanism of self-organization. A global order is organized by the simple and locally coordinated actions of autonomous agents using only local information, so complex or globally coordinated actions which use global communication and high-level strategies are not necessary. The fundamental factors for establishing a global order using self-organization are a “dissipative structure,” an “autocatalysis mechanism,” and “intentional fluctuations.” If an environment where there are agents has a dissipative structure and those agents have some sort of autocatalysis and intentional fluctuation mechanisms within themselves, it is possible to form a global order for them using only their simple and locally coordinated actions. “The blind-hunger dilemma” is used as an example to simulate the self-organization and coordinated actions of agents. In this simulation environment, there are many ant-like agents which must get energy. However, there is only one small energy supply base, so either an efficient method or the coordinated actions of agents is needed. As a result, the agents using our approach could move and get energy more efficiently than agents using conventional coordination mechanisms involving global communication and high-level strategies.

A method of gait coordination of hexapod robots using immune networks

Abstract: Biological information processing systems can be regarded as one of the ultimate decentralized systems, and have been expected to provide various fruitful ideas in the engineering field. Among these systems, the immune system plays an important role in coping with dynamically changing environments by constructing self-nonself recognition networks among different species of antibodies, and has many interesting features from an engineering stand-point, such as learning, self-organizing abilities, and so on. However, it has not yet been applied to engineering fields. Therefore we pay close attention to the immune system and attempt to construct an artificial immune network for robot control. In this study we propose a new interpretation of the roles of antibodies in terms of self-assertion and subordination, and apply this idea to a gait coordination problem of a hexapod robot as a practical example. Several computer simulations are carried out, and the robustness against disturbances and the feasibility of our method are confirmed.

Effect of complexity on learning ability of recurrent neural networks

Abstract: This paper demonstrates that recurrent neural networks can be used effectively to estimate unknown, complicated nonlinear dynamics. The emphasis of this paper is on the distinguishable properties of dynamics at the edge of chaos, i.e., between ordered behavior and chaotic behavior. We introduce new stochastic parameters, defined as combinations of standard parameters, and reveal relations between these parameters and the complexity of the network dynamics by simulation experiments. We then propose a novel learning method whose core is to keep the complexity of the network dynamics to the dynamics phase which has been distinguished using formulations of the experimental relations. In this method, the standard parameters of neurons are changed by the core part and also according to the global error measure calculated by the well-known simple back-propagation algorithm. Some simulation studies show that the core part is effective for recurrent neural network learning, and suggest the existence of excellent learning ability at the edge of chaos.

A new hybrid GA solution to combinatorial optimization problems— an application to the multiprocessor scheduling problem

Abstract: The multiprocessor scheduling problem is one of the classic examples of NP-hard combinatorial optimization problems. Several polynomial time optimization algorithms have been proposed for approximating the multiprocessor scheduling problem. In this paper, we suggest a geneticizedknowledge genetic algorithm (gkGA) as an efficient heuristic approach for solving the multiprocessor scheduling and other combinatorial optimization problems. The basic idea behind the gkGA approach is that knowledge of the heuristics to be used in the GA is also geneticized alongiside the genetic chromosomes. We start by providing four conversion schemes based on heuristics for converting chromosomes into priority lists. Through experimental evaluation, we observe that the performance of our GA based on each of these schemes is instance-dependent. However, if we simultaneously incorporate these schemes into our GA through the gkGA approach, simulation results show that the approach is not problem-dependent, and that the approach outperforms that of the previous GA. We also show the effectiveness of the gkGA approach compared with other conventional schemes through experimental evaluation.

Neural networks for control in an artificial brain of a recognition and tracking system

Abstract: This paper presents a new information-processing machine which is called the artificial brain (ABrain). It also considers the structure of artificial neural networks constructed in a Ricoh neurocomputer RN-2000 in the ABrain to track given trajectories which are produced in a micro-computer or by a light moved by hand in a recognition and tracking system.

Self-organization of complex adaptive systems as a society of rational agents

Abstract: In this paper, we aim at providing a general theoretical framework for designing complex adaptive systems as a society of rational agents. We term entities with their own interest agents. They are also rational in the sense that they only do what they want to do and what they think is in their own best interest. We formulate the dynamic interaction among those rational agents as competitive and cooperative problems. We obtain the equilibrium behavior in the long-run, and characterize the collective behavior of these rational agents. We show how complex collective behavior can emerge from the locally optimal behavior of each agent. We also describe why and how they organize themselves into a multilevel hierarchical organization with nesting structures.

Realization of artificial intelligence for integrative electroencephalogram interpretation

Abstract: A fully automatic interpretation of an awake electroencephalogram (EEG) has been developed. Automatic integrative EEG interpretation consists of four main parts: quantitative EEG interpretation, EEG report making, preprocessing of EEG data, and adaptable EEG interpretation. Automatic integrative EEG interpretation reveals essentially the same findings as those of an electroencephalographer (EEGer), and is therefore applicable in clinical situations as an additional tool for the EEGer. The method has been developed through collaboration between the engineering field (Saga University) and the medical field (Kyoto University). This work can be understood as a realization of artificial intelligence. The procedure for this realization of artificial intelligence will also be applicable in other fields of systems control.

Symmetry and symmetry-breaking in multiagent behavior

Abstract: This paper presents some results about properties of the spatial behavior of systems consisting of many artificial agents (robots). The general goal is to understand how the complexity of group behavior is related to individual behavior, and how differences arise and can be grounded. We argue that symmetry and symmetry-breaking are essential factors for the organization of spatial group behavior. They can serve as guiding principles for the construction of agents for real-world environments.

Evolution of a vision system by genetic algorithm

Abstract: This paper proposes a framework for a genetic algorithm applied to determine and construct an organ, especially the neural network of a virtual creature. The vision system of the creature is a result of genetic evolution, and we are trying to realize this on the computer. We examine how the visual organ of the animal is evolved under a special environment (e.g., the specialized visual organ of an animal to catch a moving insect), and how many variations of neural networks exist. We also think it is possible to generalize the method to an automatic generation of various kinds of visual recognition system by adding various kinds of evolution any directions.

Optimal control of active transport across a biological membrane

Abstract: We propose an optimal control principle for active transport across a biological membrane. The modeling of the membrane is based on Hill and Kedem's thermodynamic model. The performance function used to evaluate the optimality of the transport involved the rate of time-dependent changes in the concentration of particles in all the membrane layers as the state variables, and the number of receptor sites on the membrane as the control input. We decided that the optimal transport state is achieved when this cost function has been minimized under the constraints of the system equations characterizing the membrane modeling. The changes in the number of particles in the membrane layer evoked by changes in the kinetic parameters can be explained by the compensatory action of the optimal control strategy in order to prevent excessive decrease or increase of the molecular particles in all the membrane layers. The changes in the number of receptors in the paths of some physiological states can be explained by the optimal control modeling of the membrane transport. This model will be made available to create and evaluate an artificial membrane.

Robot path planning by genetic programming

Abstract: This paper describes a method to determine the path of a robot that travels around between machine tools in a production line FA factory. This decision is made by the genetic algorithm with Lisp language programming. In the algorithm, the building block method to decide fitness is adopted. The method is applied to a flexible manufacturing system (FMS) that has four machine tools and a robot.

Information processing using chaos with application to mobile robot navigation problems

Abstract: Throughout this study on information processing using an artificial neural network (ANN) and chaos we are attempting to devise a memory model that resembles human behavioral characteristics. For that purpose we construct a framework of the macroscopic model of the responding process in biological systems. Incoming stimuli are applied to the sensory receptors and preprocessed. A pattern-matching block allows one of the chaotic memories to find a feasible response in an associative way. After the chaotic memory is stabilized on one of the stable equilibrium points or limit cycles, its performance is evaluated. Since chaotic memory and the performance evaluation block form a feedback loop, they can handle features of the information blocks and store newly updated information blocks. Two kinds of chaotic memories are established in this paper: one is a 1-D map in which many information blocks can be stored as unstable periodic orbits, and the other is the famous Lozi attractor with rich dynamics. Simulations are performed for the mobile robot navigation problem in each case.