Through the integration of advanced technologies, roving robots are able to achieve high levels of robotic intelligence. Use of conventional robot programming systems to control these rovers is inappropriate due to their dynamic and not previously specified environments. As a solution to this, an expert system design is proposed that provides a viable method for the program and control of a rover. The expert system makes use of a rule based knowledge system to control the operations of a rover. A blackboard structure is utilized to facilitate communication among the various expert system components. Inference and performance logs are incorporated into the design to provide a means of verifying the expert system's performance. A prototype system has been implemented on the Ml expert system development package from Teknowledge Inc.
Vision systems capable of inspecting industrial components and assemblies have a large potential market if they can be easily programmed and produced quickly. Currently, vision application software written in conventional high-level languages such as C or Pascal are produced by experts in program design, image analysis, and process control. Applications written this way are difficult to maintain and modify. Unless other similar inspection problems can be found, the final program is essentially one-off redundant code. A general-purpose vision system targeted for the Visual Machines Ltd. C-VAS 3000 image processing workstation, is described which will make writing image analysis software accessible to the non-expert both in programming computers and image analysis. A significant reduction in the effort required to produce vision systems, will be gained through a graphically-driven interactive application generator. Finally, an Expert System will be layered on top to guide the naive user through the process of generating an application.
There have been many attempts at producing machines capable of recognising speech. Nearly all have relied on some torm of feature extraction technique followed by pattern matching using techniques such as dynamic programming. In this paper we introduce PROLOG as a language for the labelling of vowels in continuous speech. The recogniser stores descriptions of vowels which are then compared with the incoming data, where a label of a phoneme is assigned together with a value for its membership of a particular vowel class. The vowel recogniser proposed has been designed to be used together with higher level linguistic knowledge to torm the basis of a speech recognition system. The use of differing membership functions is examined.
We have implemented the development tool DIP-E, wich is based on experiences and short-comings of other similar system, as EMYCIN, OPS5 and also the programming language PROLOG. Based on modular approach in the DIP-E implementation, we have tested different inference engines for some more real problems.
An object, based on a frame representation, is designed to describe and manipulate functional properties of algorithms. This object has been defined for Image Processing in order to obtain a unified specification of Image Processing methods. It could allow to capture some expertise and to further analogy for designing new techniques. Nevertheless, its use should be appropriate in other disciplines where acquisition of some expertise is not obvious and requires efficient tools to increase operative knowledge.
The aim of this research is to create an expert system which would help with the task of identifying a biological specimen. This task is usually carried out with the aid of a set of tree-structured character descriptions called a diagnostic keg. The Dempster-Shafer theory of evidence was used to handle uncertainty associated with the date and the expertise.
This paper systematically presents a knowledge-based geological prospecting system which is intend to provide consultation on geological mapping for the aerogeophysical prospecting interpreters. The system accepts the data of an aeromagnetic survey and an aeroradioactive survey, then it uses these data to identify magnetic anomalies and radioactive anomalies, to enclose the boundaries of the anomalous bodies, and finally to distinguish the lithologies of the anomalous bodies. Yeanwhile, the system gives a confidence measure for each conclusion and answers the questions about the consultation. The capability of this system to provide the consultation relies on the knowledge the system has possessed. The knowledge has been offered by a panel of aerogeophysical pros-pecting domain experts. The system adopts a forward-backward double-threshold inexact inference strategy so that it is able to infer under the uncertain and incomplete evidences and to get approximate conclusions. In addition, the system provides explanation for its reasoning and also has knowledge acquisition functions, with those facilities the domain experts can modify and expand the knowledge base easily.
This experimental new consulting system with natural language and graphical interfaces can assist a naive user in disassem bling and reassembling a mechanical object with a cylindrical body by analyzing the structural description of the object. Many trouble shooting systems have been developed so far, but most of them do not tell us the way for decomposing the object to find out trouble points. This system is built to assist naive users in decomposing a mechanical object and in consturucting it after repair. It is difficult for a computer to give the user a series of operations necessary for exposing a trouble point by using just simple command sequences, then an integrated instruction facility using a natural language and a visual interface must be offered for specifying what portion of the object should be disassembled or constructed at the next stage, and for verifying whether what the user has done to the object is correct or incorrect.
In real world, many application areas involve rather complex problems or tasks, such as oil exploration. In developing an expert model for such kind of problem domains, abstracting and representing domain knowledge in a multilevel framework is one of efficient ways of structuring knowledge base for the systems For the last few years, we have been trying to apply Al methods, particularly expert System method to oil exploration. In this paper we will introduce SDIES, an Expert System for seismic Data Interprctation in oil exploration, with emphasis on abstraction and representation of domain knowledge.
This paper is concerned with the use of artificial intelligence technology to increase system safety in failsafe realtime systems. A safety module for a failsafe realtime system is specified which uses a production system to implement the necessary security checks. The task of this safety module is to guarantee the safety of the system. To implement the safety module production system the AI language OPS83 is used. A complete prototype for use in the Electronic Interlocking System "ELEKTRA" from ITT-Austria is being built comprising approximately 100 to 200 safety assertions in the form of production rules.
In this paper An Acoustic Imaging Recognition System (AIRS) will be introduced which is installed on an Intelligent Robotic System and can recognize different type of Hand tools' by Dynamic pattern recognition. The dynamic pattern recognition is approached by look up table method in this case, the method can save a lot of calculation time and it is practicable. The Acoustic Imaging Recognition System (AIRS) is consist of four parts -- position control unit, pulse-echo signal processing unit, pattern recognition unit and main control unit. The position control of AIRS can rotate an angle of ±5 degree Horizental and Vertical seperately, the purpose of rotation is to find the maximum reflection intensity area, from the distance, angles and intensity of the target we can decide the characteristic of this target, of course all the decision is target, of course all the decision is processed bye the main control unit. In Pulse-Echo Signal Process Unit, we ultilize the correlation method, to overcome the limitation of short burst of ultrasonic, because the Correlation system can transmit large time bandwidth signals and obtain their resolution and increased intensity through pulse compression in the correlation receiver. The output of correlator is sampled and transfer into digital data by u law coding method, and this data together with delay time T, angle information OH, eV will be sent into main control unit for further analysis. The recognition process in this paper, we use dynamic look up table method, in this method at first we shall set up serval recognition pattern table and then the new pattern scanned by Transducer array will be devided into serval stages and compare with the sampling table. The comparison is implemented by dynamic programing and Markovian process. All the hardware control signals, such as optimum delay time for correlator receiver, horizental and vertical rotation angle for transducer plate, are controlled by the Main Control Unit, the Main Control Unit also handles the pattern recognition process. The distance from the target to the transducer plate is limitted by the power and beam angle of transducer elements, in this AIRS Model, we use a narrow beam transducer and it's input voltage is 50V p-p. A RobOt equipped with AIRS can not only measure the distance from the target but also recognize a three dimensional image of target from the image lab of Robot memory. Indexitems, Accoustic System, Supersonic transducer, Dynamic programming, Look-up-table, Image process, pattern Recognition, Quad Tree, Quadappoach.
A flexible operating system using together declarative knowledge encoding, functional decomposition of applications and parallelism of a multiprocessor architecture is described. The purpose of the system is twofold: it enables to control processes in real-time, and it controls the distributed computer itself. The whole environment is described, the facilities are shown and some performance measurements are given.
This paper aims to develop a data-knowledge base system for solving decision problems which involve large amount of data under some straightforward decision rules. This data-knowledge base system can handle both data and knowledge processes. It is the expansion of relational database systems under the control of relational - DBMS. To compare the performance of this system with coventional artificial intelligence programs, a typical decision problem is programmed and operated by both dBASE III and Micro-Prolog. The study's findings suggest that this proposed system has faster executing speed for solving decision problems which have large arrays data and can represent its decision rules by normalized relations.
This paper describes the software part of a robot system developed, within the framework of the Advanced Robotic System (SYRA) project, at the GERSIC Research Laboratory. This software tool achieves the four following functions: Action , Perception , Communication. and Reasoning. The main functions of Syra are summarized in section 1. Section 2 is devoted to its knowledge representation system. The last part describes an application to the computation of trajectories in a cluttered world.
Some of the most important phases in the processing of digital images are image segmentation and classification. Contextual algorithms make use of information coming from the pictorial content (called spatial or contextual information) of the data in addition to the spectral properties. A 2-dimensional stochastic model is introduced in this paper, which enables us to combine spatial and spectral information in uniform manner. The Bayesian approach is adopted and, according to this, a pixel is assigned to that class which gives the maximum "a pos-teriori" probability conditioned on not only the observed feature vector of this pixel but also of its neighbours. Depending upon the size of the neighbourhoods and the model to compute the joint probabilities, various (formerly known or new) contextual algorithms can be derived. In order to reduce the computational complexity of the algorithms an iterative approximation is proposed and analysed. These suboptimal algorithms give considerably better results (in comparison with non-contextual ones) while time and storage requirements remain acceptable.
Digital images contain a great many samples of light intensity distribution as recorded by a vision system. Processing times using software techniques can be relatively long due to the large number of samples, and real-time operation may be impossible. The paper considers the principles of edge detection and shows how they may be applied in a real-time system. It is shown that very large scale integrated circuits may be utilised to give efficient edge detection providing the camera system operates in a non-interlaced mode.
This paper presents a multi-microprocessor implemented parallel-data driven LISP system, which can be used for expert systems, natural language machine translation and diagnostic systems. A prototype system with four LISP processors has run in HIT. A more feasible and powerful system used for Russian-Chinese machine translation is being designed and built.
Applicative models of computation based on applicative languages have, in recent years, received considerable attention as an attractive alternative to the Von Neumann model. Prolog is a high level applicative language. Its execution is conceptually an AND-OR tree search based on resolution proof procedure. This paper presents a reduction model for parallel interpretation of applicative languages such as Prolog, LISP. Reduction semantics of logic programs is analysed. A new classification of proposed models for parallel execution of logic programs is also given.
This paper presents a multi-microprocessor LISP machine whose goal is to exploit the inherent parallelism in the LISP programs fully. The base architecture is a MIMD architecture based on a hybrid model for combinating data driven, demand driven and VoN Neumann process schemes. The basic evaluation strategy is data driven. Lazy evaluation mechanism is introduced to avoid unnecessary and unsafe computations. An experimental system with the four processor elements has been built in HIT, China. The system consists of a Z80 microcomputer and three TP8O1s interconnected through three buses. Each processor evaluates a part of programs asynchronously. The shared memory is divided into two parts: list cell area and enviroment area, each of which has the indepen-dent common bus to avoid the bus bottleneck.
VISIDF is a system for generating true three-dimensional displays on flat-screened devices. Hodges and McAllister, in their article, state clearly that this system is the autostereoscopic alternative to PLZT shutter systems for computer-generated graphic appli-cations. This opens the door to consideration of the system as a component of vision for artificial intelligence applications. In order to understand the potentials of VISIDEP one must, in fact, accept several fundamental assumptions. These are: 1. Perception is an intelligent activity rather than purely stimulus/response. 2. Binocular depth cues are of greater importance to accurate depth interpretation than monocular cues. 3. Depth perception does not require object identification. Each of these assumptions is essential to the application of VISIDEP research in practical operations requiring depth interpretation. The relationships between human vision and perception and the parallax induction generated by VISIDEP technology offer depth in real time to artificial intelligence. Through machine operations on incoming data, the perception of depth is generated in much the same way as the stereoptic data enter the human being, thus providing rapidly quantifiable depth interpretation which is very accurate, perhaps more accurate that human perception of depth. The analysis of a mechanical system in relationship to human approaches to depth perceptions offers the potential of many applications of visually competent artificial intelligence. An additional factor is that the system under discussion is user friendly for human operators as well as requiring minimal reconfiguration of existing equipment and relatively simple software.
The graphical display based on a raster scan is widely applied in the technique of CAD, CAM and management, etc. It's advantage are high resolution and good quality in graphic display, low cost, convenience and flexibility in use. Hence it is rapidly developed. This article mainly discusses graphical display method, principles and various software techniques, such as control program, graphical generation software package and command form etc. based on a raster.
The paper describes a sequence of stages which permit the extraction of uniform regions and their presentation by unbroken sequences of edge points with the minimization of noise points effects on a 100 x 100 scene with 16 gray levels. Considering that the method was intended for industrial applications, it includes the elimination of background, in the first step, by employing the threshold calculated from the equalized histogram of the scene. The next step covers the filtering of quantization noise and the noise due to non-uniform illumination of working area, by fast logic clearing. The scene is then ready for extraction of contours of uniform regions, and location of residual structural noise due to inadequate resolution of camera of an emphatic discontinuity in illumination (Ex. shining objects). The final step covers the scene scanning by means of adaptive masks for elimination of structural noise and follower coding of uniform regions in the form of sequences of sucessive connected points. The scene coded in such a method comprizes all information about objects on it, and permits their recovery by employing pattern recognition algorithms.
As a natural development of work concerning the measurement of depth or range in three-dimensional television displays a prototype robot vision system has been investigated. The system uses a structured lighting technique and newly developed photogrammetric algorithms. The system gives a realistic cycle time for the acquisition of a complete coordinated matrix of image points. The pitch of the matrix could be varied for coarse or fine resolution and a measurement window could be defined anywhere in the field of view of the stereoscopic camera.
In this paper, a new digital image texture analysis method, i.e., gray level and energy cooccurrence image texture analysis method, is presented. According to the method, the two image texture analysis models are designed. They are 3 X 3 no-zero-sum mask model and gray level-difference cooccurrence matrix model. And their texture parameters are calculated by using these models, separately. The satisfactory test results are acquired for classifying the sensing images.
This paper describes an octant structure for three-dimensional objects. A direct proce-dure for encoding larger octant belonging to an object is presented. Operators associated with octants are provided. Union, intersection, and complement of octants encoded objects are computed. Finally we develop data structures for octant operating.
A smart sensor system for real time remote sensing image processing on-board satellites is described. In order to implement the optimum system, its hardware and software should be taken an overall consideration. A serial pipeline suited to scanning data processing in real time is adopted as the system architecture. Meanwhile, 3 X 3 mask texture analysis and Hadamard-DPCM data compression suited to the pipeline processing are selected as the system algorithms. Based on the experiment of the model, a practical smart sensor system is presented.
We present a method to model industrial parts with generalized cones. These cones are defined by their bottom, their top, and the length of the spine. Bottom and top may be rectangles, circles, rings, or other easily described planar surfaces. Objects are built by the union of such cones. To use these models in image analysis you can give visible features to the surfaces of the cones and you can define the parameters of the cones with fuzzy functions. The modeling system is built up in a relational database with relations of first normal form.
In the fields of computer vision and scene analysis, it is difficult to recognize a 3-dimensional object from 2-dimensional image taken by a TV camera, because the image changes as the object is rotated in 3-dimensional space. This paper deals with convex polyhedrons, and proposes a method for determining 2-dimensional visible patterns of a given 3-dimensional object. Each visible pattern is defined by the shapes of visible faces of the 3-dimensional object, and by the connection relationships between the faces. This method makes it easier to recognize a 3-dimensional object from 2-dimensional images. Finally we present some experimental results to show the effectiveness of our method.
Image segmentation by texture is a classical problem of digital image analysis. Recently, the author has formulated a new segmentation problem which is based on the notion of texture defect, or imperfection. Imperfections are small non-textured regions that locally break the homogeneity of the texture pattern. Detection of imperfections is a novel, specific task differing from the classical texture segmentation. Imperfection analysis provides a necessary basis for designing particular application techniques aimed at de-fect detection in various fields of visual inspection of surfaces, such as indication of defects in wood or textiles. In this paper, the main principles of imperfection analysis are briefly outlined and a few dedicated inspection algorithms mentioned. A new procedure is suggested for the adaptive detection of texture imperfections in the presence of a slow spatial variation of the texture pattern. Key words : texture segmentation, surface inspection, texture imperfections.
Current trends in knowledge-based computing have produced a large number of expert system building tools. This onslaught of high-tech software stems from the discovery that expert systems can be effectively applied to a variety of industrial and military problem domains. A variety of vendors provide expert system prototyping and development tools which greatly accelerate the construction of intelligent software. Today's expert system tool generally provides the user with a friendly interface, an efficient inference engine, and formalisms that simplify the creation of a domain knowledge base. This paper presents a formalism for expert system tool evaluation and critiques an exhaustive variety of commercially available tools.