Showing papers in "Artificial Life and Robotics in 2000"

Evolving robot morphology and control

Abstract: Most robotic approaches beging with a fixed robot hardware design and then experiment with control structures. We take a different approach that considers both the robot hardware and the control structure as variables in the evolution process. This paper reports the results of experiments which explore the placement of sensors and effectors around the perimeter of a simulated agent's body, and the neural network (NNet) that controls them.

An overview of the development of Handy 1, a rehabilitation robot to assist the severely disabled

Abstract: The Handy 1 was developed in 1987 by Mike Topping to assist an 11-year-old boy with cerebral palsy to eat unaided. The system is the most successful low-cost, commercially available robotic system in the world to date, and is capable of assisting the most severely disabled with several everyday functions such as eating, drinking, washing, teeth-cleaning, shaving, and make-up application. This paper charts the development of the system, gives an overview of some individual Handy 1 studies, and also looks at the development of the robotic aid to independent living (RAIL) system funded by the European Commission BIOMED II program.

Walking-support robot system for walking rehabilitation: design and control

Abstract: An intelligent walking-assistance robot system has been developed to help the elderly or the disabled in rehabilitation programs. From the design viewpoint, several different mechanisms to satisfy the strict requirements for use in a rehabilitation program were considered and studied. A two-wheel mobile mechanism was developed to provide motions to follow the patient's walking trajectory, and also to make the patient follow a specified trajectory. Equations of motion were derived for the unloading control, and a force control algorithm was developed. For motion control of the mobile base, virtual trajectory planning by the B-spline method and PID control were used. Sensing the motion of the patient is performed by a linear potentiometer and a rotating encoder attached to the robot manipulator. The system was tested on patients in hospital and the experimental results are reported.

Design of an auditory guidance system for the blind with signal transformation from stereo ultrasonic to binaural audio

Abstract: We have designed an auditory guidance system for the blind using ultrasonic-to-audio signal transformation. We first investigated the system requirements, and designed a simple but useful portable guidance system for the blind. The system derives visual information using multiple ultrasonic sensors, and transforms it to binaural auditory information using a suitable technique. The user can recognize the position of obstacles and the surrounding environment. The system is composed of two parts. One is a glasses-type system, and the other is a cane-type system with guide wheels. The former functions as an environment sensor, and the latter functions as a clear-path indicator. Wide-beam-angle ultrasonic sensors are used to detect bojects over a broader range. The system is designed as a battery-supplied portable model. Our design is focused on low power consumption, small size, light weight, and easy manipulation.

Winning strategies for robotic wars: defense applications of linguistic geometry

Abstract: This paper reports new results of research, which started in 1972 in Moscow, USSR. For 16 years Boris Stilman was involved in the advanced research project PIONEER led by a formed World Chess champion, Professor Mikhail Botvinnik. The goal of the project was, at first, to discover and mathematically formalize methodologies utilized by the most advanced chess experts (including Botvinnik himself) in solving chess problems almost without search. The next step was to apply this new theory to complex search problems from various problem domains. In the 1980s, in Moscow, Stilman developed the foundations of the new approach. In 1991, while at McGill University, Montreal, Canada, Stilman coined the term “Linguistic Geometry” (LG) as a name for the new theory for solving Abstract Board Games. After 1991, this research continued at the University of Colorado at Denver, USA. In 1995, V. Yakhnis joined the LG effort. In the 1990s, it was shown that LG is applicable to a wide class of higher-dimensional, multi-agent games with concurrently moving agents, which are ideally suited for combat planning and control. Also, it was proved that for several classes of games LG generates optimal strategies in polynomial time. This groundbreaking results also suggests that for much wider classes of games LG strategies are also optimal or close to optimal. Over a hundred papers on LG have been published. Stilman wrote the first scholarly book on LG,Linguistic Geometry: From Search to Construction, published in February 2000. Over the last two years, defense applications of LG have attracted so much attention at the Defense Advanced Research Projects Agency (DARPA), Rockwell, and Boeing that the number of LG-based projects has skyrocketed. In 1999, recognizing the maturity and power of this technology, a group of scientists, engineers, and entrepreneurs founded a company, STILMAN Advanced Strategies, to facilitate development of government and commercial applications of LG.

Wearable master device for spinal injured persons as a control device for motorized wheelchairs

Abstract: This paper describes a wearable, master device for people with a spinal injury who can move their neck and shoulders but cannot move their legs and arms. A device that measures the movements of their neck or shoulder can help them to drive a wheelchair. The sensors of such a wearable master device must be lightweight, small, and easily attached to cloth. Therefore, optical fiber curvature sensors are used to measure the human body motion. For a previously developed wearable master device, two calibration and mapping methods with, the sensors are proposed to extract 2-DOF human shoulder motions. One is constructed with simple geometric equations. The other is constructed with a multilayered artificial neural network. The two methods are compared. Experimental results show that the wearable master device can be used effectively for a 2-DOF input device for handicapped persons. It was also shown that a subject can control a mobile robot with the wearable master device.

Improved genetic algorithm for generalized transportation problem

Abstract: In this paper, we introduce the genetic algorithm approach to the generalized transportation problem (GTP) and GTP with a fixed charge (fc-GTP) We focus on the use of Prufer number encoding based on a spanning tree, which is adopted because it is capable of equally and uniquely representing all possible trees From this point, we also design the criteria by which chromosomes can always be converted to a GTP tree The genetic crossover and mutation operators are designed to correspond to the genetic representations With the spanning-tree-based genetic algorithm, less memory space will be used than in the matrix-based genetic algorithm for solving the problem; thereby computing time will also be saved In order to improve the efficiency of the genetic algorithm, we use the reduced cost for the optimality of a solution and the genetic algorithm to avoid degeneration of the evolutionary process A comparison of results of numerical experiments between the matrix-based genetic algorithm and the spanning-tree-based genetic algorithm for solving GTP and fc-GTP problems is given

The degree of human visual attention in the visual search

Abstract: Human beings can obtain visual information in parallel through the retina, but they cannot pay attention to all the information at the same time. In psychological studies, the human characteristics of visual attention have often been investigated by analyzing the characteristics of the visual search task. Previous studies suggested that the information features of the visual search task are processed in parallel at early stages of processing. However, the authors consider that these features are not processed completely in parallel, and have a reciprocal action to each other. In order to clarify the reciprocal action of the features in a visual search and the continuity of visual attention, the characteristics of reaction times were measured with changing forms of visual stimuli. The experimental results suggested that the reaction time changed when the features of the visual stimuli in the visual search task changed. This means that the features are affected by each other. Furthermore, continuity of reciprocal action is also suggested, and the degree of visual attention is decided by this continuity. The results provided significant basic data to support our proposed mathematical model of visual attention.

Functional electrical stimulation

Abstract: Percutaneous intramuscular electrodes and a portable multichannel system were used to restore the function of paralyzed upper and lower extremities in spinal cord injuries and hemiplegic patients. The hybrid functional electrical stimulation (FEC) using percutaneous intramuscular electrodes provides practical ambulation for paraplegics. FES was more effective than the flexor hinge splint in increasing the grasping power (GP) of tetraplegic patients, and a stronger and stable GP which was not affected by wrist position make FES practical for improving the activities of daily living. A tilt sensor, which was put on the thigh, could be used to measure a gait cycle. This study suggests that putting a tilt sensor on a hemiplegic patient's thigh improves the FES procedure and may help to restore gait in these patients. We conclude that percutaneous intramuscular FES is useful for spinal cord injuries and hemiplegic patients.

BizSim: the world of business—in a box

Abstract: Large-scale, agent-based simulations are dramatically changing how one studies real-world processes, especially social and behavioral processes. This paper addresses some of the ways this computing technology enables us to understand more about the world of business. In aprticular, we show here how computer simulation has been used to investigate the efficiency of stock markets and the operation of a supermarket. Both examples illustrate the point that it is possible to gain a much deeper understanding of the ways of business by screating “laboratories” for doing controlled, repeatable scientific experiments with such systems.

Evolving complex group behaviors using genetic programming with fitness switching

Abstract: Genetic programming provides a useful tool for emergent computation and artificial life research. However, conventional genetic programming is not efficient enough to solve realistic multiagent tasks consisting of several emergent behaviors that need to be coordinated in the proper sequence. In this paper, we describe a novel method, called fitness switching, for evolving composite cooperative behaviors in multiple robotic agents using genetic programming. The method maintains a pool of basis fitness functions which are switched from simpler ones to more complex ones. The performance is demonstrated and evaluated in the context of a table transport problem. Experimental results show that the fitness switching method is an effective mechanism for evolving collective behaviors which can not be solved by simple genetic programming.

Artificial realization of human on-off decision making based on the conditional probability of a database

Abstract: As human on-off decisions are the basic problems in our human lives, the analysis of human on-off decision making is an interesting topic. The procedures of qualified human decision making include many intuitive factors which have been acquired from previous valuable experience and gained through learning, but they may not be easily understood by others within a short period. By the use of a database of causes and decisions made by qualified experts for an objective event, human decision making for that event can be realizable artificially. This paper investigates a general method for realizing artificial human on-off decision making based on the conditional probability of the database. As on-off decision making is a discrete event and the causes for that decision making are continuous events, a mathematical treatment of a Dirac delta function in a probability density function is required to derive the conditional probability for the decision making. Several examples of artificial human decision making by the proposed method were demonstrated, and the results obtained showed good agreement with those of human experts in the respective fields.

Cooperative exploration of distributed worlds in WAVE

Abstract: This paper modifies the WAVE model for parallel processing in virtual networks to explore and process continuous physical worlds. Expressing distributed activity by cooperative jobs spreading in space and “seeing” each other, rather than as vehicles exchanging messages, the model allows complex missions to be planned in a very flexible manner, with mobile hardware being assigned to the evolving space-conquering programs (waves) dynamically, when required or available. A number of cooperative scenarios in a physical world demonstrate the simplicity and compactness of the wave code. The execution of waves by mobile hardware is discussed, including run-time mapping of waves to vehicles, and supporting multiple distributed jobs in cases of hardware shortages. WAVE can be used efficiently for solving complex problems in space by organized groups of cheap specialized mobile robots, where intelligent behavior is provided by very high level of system organization rather than by the smartness of individual units. In a broader sense, it may also serve as a basic technology for parallel and distributed simulation, and the management of evolution and self-organization of large open systems of different natures.

Speech recognition using a stereo vision neural network model

Abstract: This paper describes a new algorithm for speech recognition by using stereo vision pattern recognition equations with competition and cooperation. In our research, we applied recently developed 3-layered neural net (3LNN) equations to speech recognition. Our proposed acoustic models using these equations yield better recognition results than the hidden Markov model (HMM). When using a 216 (240) word database, stereo vision acoustic models gave 6.5% (6.6%) higher accuracy than HMMs.

Information transformation by virus-evolutionary genetic programming

Abstract: This paper deals with genetic programming (GP) for information translation. The GP can generate a structured computer program, but it is difficult to define recursive functions automatically. Therefore, we propose a virus-evolutionary genetic programming (VE-GP) composed of two populations; host and virus. Here, a virus plays the role of an automatically defined function. First, the VE-GP is applied to a function approximation problem, and the simulation result shows that the VE-GP can generate a function to approximate the given function with small errors. Next, the VE-GP is applied to the information transformation for a classification task, and the simulation result shows that the VE-GP can generate a function to classify a given data set.

Immersive user environments in rehabilitation robotics and mechatronics

Abstract: Advances in software, computing, control architectures, and design practice have led to rehabilitation robots with high-bandwidth, bidirectional user interfaces that allow people to interact with mechatronic systems in highly realistic, cognitively and biomechanically coupled scenarios. Immersive and graphic interfaces to real-time systems such as robots require, robust control environments and high-speed computing to provide safe, human-scale motions and interfaces with sufficient quality to be usable in functional environments. This paper illustrates some emerging applications in rehabilitation, spanning assistive technology, simulation/design aids, and smart therapy devices. Examples will be drawn laregly from work done at the VA Palo Alto Rehabilitation R&D Center.

A fuzzy model to control the temperature in the cooling of metal molds with a spray robot

Abstract: In a die-casting operation, the die spray is the key to quality and productivity, but the die-spray operation depends on parameter settings such as the spray volume and the operator's experience, even when a robot sprayer is used. The purpose of this study was to establish an intelligent system to control the surface temperature of a die with a water-based mist spray. A fuzzy model was applied to determine the optimum mist spray volume. The results generated show a good match with the actual measurements of the rate of the temperature drop versus the spray-mist volume. After the fuzzy analysis, the spray parameters calculated were used to control an actual die temperature experiment conducted on a heated plate model. The fuzzy model constructed by using fundamental experiments allowed the die spray operation to maintain a predetermined spray volume.

Some aspects of human-friendly control for movement-helping devices

Abstract: Future welfare models should meet the needs of a large group of aged and disabled people. Present research and technology efforts in this direction are oriented toward the development of highly effective home-installed devices which will provide the user with continuous assistance in different everyday activities. The problems and needs of the elderly and the disabled vary from person to person over a large range, and with many serious requirements. Modern designs for such devices are oriented toward solutions where the user is less engaged in the control process, and the interaction between the device and the user ishuman-friendly, i.e., the devices should possess a high level of intelligence in their controls, actions, and interactions with the user, offering him/her a high level of comfort and functionality. This paper considers a mixed control mode as a better alternative for human-machine interactions for users with great movement disability. Some examples of easy direct control are discussed.

On Pattern Classification of EMG Signals for Walking Motions

Abstract: We present a method to calssify electromyogram (EMG) signals which are utilized as control signals for a patient-responsive walker-supported system for paraplegics. Patterns of EMG signals for different walking motions are classified via adequate filtering, real EMG signal extraction, AR-modeling, and a modified self-organizing feature map (MSOFM). In particular, a data-reducing extraction algorithm is employed for real EMG signals. Moreover, MSOFM classifies and determines the results automatically using a fixed map. Finally, the experimental results are presented for validation.

Ola—what goes up must fall down: emergent Mexican wave behavior

Abstract: We made a LEGO Mindstorms robot soccer model using a distributed behaviour-based system, which was showcased at RoboCup98 during the soccer World Cup in France 1998. To put the robot soccer game into an appropriate context, we also constructed a stadium out of LEGO pieces, including stadium lights, rolling commercials, moving cameras projecting images onto big screens, a scoreboard and approximately 1500 small LEGO spectators who did the “Mexican wave” as seen in soccer stadiums. These devices were controlled using the LEGO Dacta Control Lab system and the LEGO CodePilot system that allow programming motor reactions which can be based on sensor inputs. The wave of the LEGO spectators was made using the principle ofemergent behaviour. There was no central control of the wave, but it emerges from the interaction between small units of spectators with a local feedback control.

Generation of jumping motion pattern for a hopping robot using genetic algorithm

Abstract: This paper describes an application of genetic algorithm to generate a jumping motion pattern for a hopping robot. A central pattern generator is used to generate the motion pattern. The tuning parameters of the central pattern generator are regarded as genes and adjusted by the genetic algorithm, so that the hopping robot can jump continuously to the reference height with the minimum force. To realize online tuning of the parameters, new genetic operations such as few individuals, quick estimation, instant selection, and intentional mutation are introduced. The experimental results demonstrate the effectiveness of the proposed scheme.

A new learning method using prior information of neural networks

Abstract: In this paper, we present a new learning method using prior information for three-layer neural networks. Usually when neural networks are used for identification of systems, all of their weights are trained independently, without considering interrelated weights values. Thus, the training results are usually not good. The reason for this in that each parameter has its influence on others during learning. To overcome this problem, we first give an exact mathematical equation that describes the relation between weight values given a set of data conveying prior information. The we present a new learning method that trains part of the weights and calculates the others using these exact mathematical equations. This method often a priori keeps the given mathematical structure exactly the same during learning; in other words, training is done so that the network follows a predetermined trajectory. Numerical computer simulation results are provided to support this approach.

Learning algorithms in a class of knowledge-based systems

Abstract: A control system with a static plant described by knowledge representation in the forms of relations and of logical formulas is considered. The learning process consists in using successive knowledge validation and updating it to make current control decisions. Two approaches and algorithms are described: one for the validation and updating of knowledge about the plant, and one on the form of control. In both cases two versions are presented: the learning process in either an open-loop or a closed-loop control system. For a plant with logical knowledge representation, the logic-algebraic method is applied.

Intelligent control of a robot manipulator by visual feedback

Abstract: We present a new approach to visual feedback control using image-based visual servoing with stereo vision. In order to control the position and orientation of a robot with respect to an object, a new technique is proposed using binocular stereo vision. The stereo vision enables us to calculate an exact image Jacobian not only at around a desired location, but also at other locations. The suggested technique can guide a robot manipulator to the desired location without needing such a priori knowledge as the relative distance to the desired location or a model of the object, even if the initial positioning error is large. We describes a model of stereo vision and how to generate feedback commands. The performance of the proposed visual servoing system is illustrated by a simulation and by experimental results, and compared with the conventional method for an assembling robot.

Generating novel memories by integration of chaotic neural network modules

Abstract: A principle of integrating neural network modules based on chaotic dynamics was studied on our two-moduled Nozawa model. Chaotic neural networks represent each embedded pattern as a low-dimensional periodic orbit, and the others are shown as high-dimensional chaotic attractors. This is equivalent to W. Freeman’s “I don’t know” and “I know” states. In particular, we noted that the combination of two-way inputs to each neural network module conflicted with embedded Hebbian correspondence. It was found that the interaction between the modules generated a novel “I know” state in addition to the embedded representation. Chaotic neural network modules can autonomously generate novel memories or functions by this interaction. The result suggests a functional integration in neural networks as it ought to be, e.g., feature binding and gestalt.

A polynomial fuzzy neural network for modeling and control

Abstract: This paper introduces a novel neurofuzzy system based on polynomial fuzzy neural network (PFNN) architecture. A PFNN consists of a set of if-then rules with appropriate membership functions (MFs) whose parameters are optimized via a hybrid genetic algorithm. A polynomial neural network is employed in the defuzzification scheme to improve output performance and to select appropriate rules. A performance criterion for model selection is defined to overcome the overfitting problem in the modeling procedure. For a performance assessment of the PFNN inference system, two well-known problems are employed for a comparison with other methods. The results of these comparisons show that the PFNN inference system out-performs the other methods and exhibits robustness characteristics.

An adaptive classifier system tree for extending genetics-based machine learning in a dynamic environment

Abstract: An autonomous agent should possess the ability to adapt its cognition structure to a dynamically changing environment. This ability may be achieved when au- tonomous agents interact with the environment. In this paper, an adaptive classifier system tree is proposed for extending genetics-based machine learning in a dynamic environment. The architecture has the properties of self- similarity and self-organization. When environmental changes are inspected, the autonomous agent can adapt its cognition structure to the new environment so that cognition can be achieved with great efficiency. After description of the dynamic structure and the principle of the structure's self-organization, some experiments illustrating how the architecture works are described and discussed.

Assistant force compensation for hand movement of patients by using exogenous signals and/or neurophysiological signals

Abstract: Because functional diseases of the brain can cause disabilities related to human movement control, a compensation method was developed for improving the performance of hand movements. The compensation for human hand movements can be carried out by adding an assistant force that is generated from artificial equipment attached to a human arm. From the experiment on visual target tracking, it was found that the tracking trajectory was adequately represented by a dynamic model of the motion of an articulated industrial robot arm, and the different abilities for movement control among healthy people and patients were classified by different model parameters as position loop gain, velocity loop gain, and response delay. Dynamic force compensation was approached by considering the different control features of the patients. The effectiveness of the proposed compensation method was verified in a simulation study on an actual industrial robot arm. A human-machine interface, e.g., a brain-computer interface (BCI), for realizing the control of artificial equipment to compensate for human hand movements is also presented and discussed.

An adaptive associative memory system based on autonomous reaction between image memories

Abstract: To construct a “thinking-like” processing system, a new architecture of an adaptive associative memory system is proposed. This memory system treats “images” as basic units of information, and adapts to the environment of the external world by means of autonomous reactions between the images. The images do not have to be clear, distinct symbols or patterns; they can be ambiguous, indistinct symbols or patterns as well. This memory system is a kind of neural network made up of nodes and links called a localist spreading activation network. Each node holds one image in a localist manner. Images in high-activity nodes interact autonomously and generate new images and links. By this reaction between images, various forms of images are generated automatically under constraints of links with adjacent nodes. In this system, three simple image reaction operations are defined. Each operation generates a new image by combining pseudofigures or features and links of two images.

Intelligent fault-tolerant system of vibration control for flexible structures

Abstract: This paper describes an intelligent fault-tolerant control method for vibration control of flexible structures. We consider a case where the fault phenomena of the control system for flexible structures can be treated as a change of system parameters. Therefore, the adaptive control method can be applied to a vibration control system for flexible structures with a fault. In this paper, a neural network (NN) adaptive control system is used to compensate for the change in the parameters of a plant with a fault. When the characteristics of the plant and of a nominal model have been agreed by a NN adaptive control system, the control method designed for the nominal model, such as decoupling feedback control or linearizing feedback control, can be used even if the change in the system parameters has been caused by a fault. To confirm the effectiveness of the proposed fault-tolerant control method, the simulational results from a 5-link robotic arm are shown at the end of the paper.