We propose the use of cubic quadratic cubic squared (CQCS) spline for the trajectory generation in Cartesian space. Use of CQCS spline gives simple analytical solution to minimum time trajectory generation with velocity and acceleration constraints. The expressions for wandering time and wandering acceleration are also calculated. A straight line path with constant maximum allowed speed in minimum time can be generated with this method. This property leads to interpolate two position points by constant speed straight line motion with smooth transition. The advantage of this method is that the trajectory thus obtained is traversed in minimum time while passing through the given intermediate points. The simplicity of this method makes its on-line computation possible.

CONTROLAB is an environment which integrates intelligent systems and control algorithms aiming at applications in the area of robotics. Within CONTROLAB, two neural network architectures based on the backpropagation and the recursive models are proposed for the implementation of a robust speaker-independent word recognition system. The robustness of the system using the backpropagation network has been largely verified through use by children and adults in totally uncontrolled environments such as large public halls for the exhibition of new technology products. Experimental results with the recursive network show that it is able to overcome the backpropagation network major drawback, the frequent generation of false alarms. In addition, within CONTROLAB, the trajectory to be followed by a robot arm under self-tuning control is determined by a system which uses either VGRAPH or PFIELD algorithms to avoid obstacles detected by the computer vision system. The performance of the second algorithm is greatly improved when it is applied under the control of a rule-based system. An application in which a SCARA robot arm is commanded by voice to pick up a specific tool placed on a table among other tools and obstacles is currently running. This application is used to evaluate the performance of each sub-system within CONTROLAB.

The teaching and playback method is mainly a teaching technique for industrial robots. However, this technique takes time and effort in order to teach. In this study, a new teaching algorithm using stereo vision based on human demonstrations in front of two cameras is proposed. In the proposed teaching algorithm, a robot is controlled repetitively according to angles determined by the fuzzy sets theory until it reaches an instructed teaching point, which is relayed through cameras by an operator. The angles are recorded and used later in playback. The major advantage of this algorithm is that no calibrations are needed. This is because the fuzzy sets theory, which is able to express qualitatively the control commands to the robot, is used instead of conventional kinematic equations. Thus, a simple and easy teaching operation is realized with this teaching algorithm. Simulations and experiments have been performed on the proposed teaching system, and data from testing has confirmed the usefulness of our design.

An open architecture framework for intelligent multisensor integration in an industrial environment is being developed. This framework allows for the computational evaluation and understanding of sensor uncertainty and data validity through the comparison of sensor data in a common format. A logical sensor model is used to represent both real and abstract sensors within the architecture. This allows for the unobtrusive addition or replacement of sensors. All logical sensor outputs are accompanied by a corresponding confidence level. These confidences are used to dynamically allocate valid sensor readings for use by higher-level sensors. Sensory information is passed to an inference engine which uses user- selectable and adjustable fuzzy logic and/or neural network modules to provide the required decision making intelligence. This architecture may be applied to a broad range of industrial applications, especially those involving non- uniform product grading.

Ultrasonic is a popular nondestructive testing technique for detecting flaws in metals, composites and other materials. A major limitation of this technique for successful field implementation is the need for skilled labor to identify an appropriate testing methodology and conduct the inspection. A knowledge-based assistant that can help the inspector in choosing the suitable testing methodology would greatly reduce the cost for inspection while maintaining reliability. Therefore a rule-based decision logic that can incorporate the expertise of a skilled operator for choosing a suitable ultrasonic configuration and testing procedure for a given application is explored and reported in this paper. A personal computer (PC) based expert system shell, VP Expert, is used to encode the rules and assemble the knowledge to address the different methods in ultrasonic inspection for metals. The expert system will be configured in a question-answer format. Since several factors (such as frequency, couplant, sensors, etc.) influence the inspection, appropriate decisions have to be made about each factor depending on the type of inspection method and the intended use of the metal. This knowledge base will help in identifying the methodology for detecting flaws, cracks, and thickness measurements, etc., which will lead to increase safety.

On-line inspection of PCBs requires acquisition and processing of gigabytes of image data in a matter of few seconds, especially when multi-layer and very high-resolution boards are used. To meet the demands for speed and accuracy, our inspection system uses run-length encoding (RLE) for storage and operations and an inspection scheme which exploits the availability of an artwork for comparison purposes. The system which is suitable for parallel processing consists of four parts: (1) segmentation of artwork and feature extraction, (2) image acquisition, (3) inspection of blank areas, and (4) inspection of trace areas. First, the artwork which is available as a CAD file in Gerber format, is segmented offline into primitive patterns and information related to the location and identification of each segment is stored in a large image database to be used later for real-time inspection. A time- and space-efficient technique based on RLE is used for storage of essential features and image operations.

The total system throughput (ST) is one of the most important decision variables at the planning/scheduling phase of a manufacturing system. Material requirement planning (MRP) and master production schedule (MPS) are based on the assumption that ST is known. All the subsequent developments (e.g. jobs- release, system work-load, input-product mix, etc.) depends on such an assumption. If this assumption is incorrect, the production activity control (PAC) will not be able to satisfy the planned targets during the scheduling phase. Delays and bottlenecks will be unavoidable in the system. In case of random flexible manufacturing system (FMS) (or, in general, job-shop production), the measure of ST can not be evaluated a priori without running simulations or observing the actual flow of the operations in the system. The way entities enter the system (sequencing and percentage of the input products) effects the value of ST in such a way that estimation based on historical data are highly risky. The methodology proposed in this paper allows the scheduler to assess the analytical functions which link ST, and other output performance variables, to the input product mix (IPM). This way the robustness of the scheduling plan can be verified before the actual release of the jobs into the system.

Automated product assembly systems are traditionally designed with the intent that they will be operated with few significant changes for as long as the product is being manufactured. This approach to factory design and programming has may undesirable qualities which have motivated the development of more 'flexible' systems. In an effort to improve agility, different types of flexibility have been integrated into factory designs. Specifically, automated assembly systems have been endowed with the ability to assemble differing products by means of computer-controlled robots, and to accommodate variations in parts locations and dimensions by means of sensing. The product life cycle (PLC) is a standard four-stage model of the performance of a product from the time that it is first introduced in the marketplace until the time that it is discontinued. Manufacturers can improve their return on investment by adapting the production process to the PLC. We are developing two concepts to enable manufacturers to more readily achieve this goal: the agile assembly architecture (AAA), an abstract framework for distributed modular automation; and minifactory, our physical instantation of this architecture for the assembly of precision electro-mechanical devices. By examining the requirements which each PLC stage places upon the production system, we identify characteristics of factory design and programming which are appropriate for that stage. As the product transitions from one stage to the next, the factory design and programing should also transition from one embodiment to the next in order to achieve the best return on investment. Modularity of the factory components, highly flexible product transport mechanisms, and a high level of distributed intelligence are key characteristics of minifactory that enable this adaptation.

This paper discusses the problem of integrating human intelligence and skills into an intelligent manufacturing system. Our center has jointed the Holonic Manufacturing Systems (HMS) Project, an international consortium dedicated to developing holonic systems technologies. One of our contributions to this effort is in Work Package 6: flexible human integration. This paper focuses on one activity, namely, human integration into motion guidance and coordination. Much research on intelligent systems focuses on creating totally autonomous agents. At the Center for Intelligent Systems (CIS), we design robots that interact directly with a human user. We focus on using the natural intelligence of the user to simplify the design of a robotic system. The problem is finding ways for the user to interact with the robot that are efficient and comfortable for the user. Manufacturing applications impose the additional constraint that the manufacturing process should not be disturbed; that is, frequent interacting with the user could degrade real-time performance. Our research in human-robot interaction is based on a concept called human directed local autonomy (HuDL). Under this paradigm, the intelligent agent selects and executes a behavior or skill, based upon directions from a human user. The user interacts with the robot via speech, gestures, or other media. Our control software is based on the intelligent machine architecture (IMA), an object-oriented architecture which facilitates cooperation and communication among intelligent agents. In this paper we describe our research testbed, a dual-arm humanoid robot and human user, and the use of this testbed for a human directed sorting task. We also discuss some proposed experiments for evaluating the integration of the human into the robot system. At the time of this writing, the experiments have not been completed.

This work has been developed within the framework of agent- based decentralized scheduling for flexible manufacturing systems. In this framework, all workcells comprising the manufacturing system, and the products to be generated, are modeled via intelligent software agents. These agents interact dynamically using a bidding production reservation (BPRS) scheme, based on the Contract Net Protocol, to devise the production schedule for each product unit. Simulation studies of a job shop have demonstrated the gains in performance achieved by this approach over heuristic dispatching rules commonly used in industry. Manufacturing environments are also prone to operational uncertainties such as variations in processing times and machine breakdowns. In order to cope with these uncertainties, the BPRS algorithm has been extended for dynamic rescheduling to also occur in a fully decentralized manner. The resulting multi-agent rescheduling scheme results in decentralized control of flexible manufacturing systems that are capable of responding dynamically to such operational uncertainties, thereby enhancing the robustness and fault tolerance of the proposed scheduling approach. This paper also presents the effects of the proposed agent-based decentralized scheduling approach on the performance of the underlying flexible manufacturing system under a variety of production and scheduling scenarios, including forward and backward scheduling. Future directions for this work include applying the proposed scheduling approach to other advanced manufacturing areas such as agile and holonic manufacturing.

Reduced time to market places exacting requirements for the speed and quality of the design, manufacture and testing of new plastic parts. Traditionally, the greatest time for getting a new plastic product prototyped and tested is that for the mold fabrication phase. New metal materials for selective laser sintering (SLS) rapid prototyping technology allow direct mold fabrication for prototype plastic parts. Typically these molds are also useable for small scale production runs up to 50,000 parts. Using this technology prototype parts can be manufactured using the same materials and processes as used for the final product allowing testing of the whole manufacturing process for the prototype. This gives a company new opportunities to get a new or modified product to market faster and cheaper than by using traditional mold making processes. In this paper we describe this new technology and discuss how small and mid-size manufacturing companies can benefit from it.

Manufacturing is currently undergoing a revolutionary transition with focus shifting from mass production to mass customization. This trend motivates a new generation of advanced manufacturing systems that can dynamically respond to customer orders and changing production environments. It is becoming increasingly important to develop control architectures that are modifiable, extensible, reconfigurable, adaptable, and fault tolerant. Heterarchical control structures, made up of multiple, distributed, locally autonomous entities, provide this kind of control. Our research focus is on efficient and effective scheduling and routing methodologies that can be applied to heterarchically controlled manufacturing processes. The Contract-Net based scheduling approach, developed in distributed artificial intelligence (DAI), adopts a multi-agent cooperative problem- solving paradigm based on bidding and negotiation mechanisms to implement production plans as distributed and localized schedules for individual workstations. This paper discusses a Contract-Net based scheduling algorithm in a realistic manufacturing testbed, a model induction motor assembly plant. This testbed, developed as part of the HMS project, is a typical example of low-volume, high-variety production facility, and it highlights many of the problems that arise from the inflexibility of centralized management system architectures.

The life cycles of products have been getting shorter. To meet this rapid turnover, manufacturing systems must be frequently changed as well. In engineering to develop manufacturing systems, there are several tasks such as process planning, layout design, programming, and final testing using actual machines. This development of manufacturing systems takes a long time and is expensive. To aid the above engineering process, we have developed the virtual manufacturing workcell (VMW). This paper describes a concept of VMW and design method through computer aided manufacturing engineering using VMW (CAME-VMW) related to the above engineering tasks. The VMW has all design data, and realizes a behavior of equipment and devices using a simulator. The simulator has logical and physical functionality. The one simulates a sequence control and the other simulates motion control, shape movement in 3D space. The simulator can execute the same control software made for actual machines. Therefore we can verify the behavior precisely before the manufacturing workcell will be constructed. The VMW creates engineering work space for several engineers and offers debugging tools such as virtual equipment and virtual controllers. We applied this VMW to development of a transfer workcell for vaporization machine in actual manufacturing system to produce plasma display panel (PDP) workcell and confirmed its effectiveness.

This paper presents an approach to extract manufacturing features and their corresponding attributes from a constructive solid geometry based commercial CAD software. Manufacturing features are limited to blind hole, through hole, blind pocket, through pocket, depression, notch, slot, and step. The attributes are limited to height, length, and width for a block feature, height and diameter for a hole feature.

In this paper we deal with the last control development technique implemented at the PERCRO. We show an integrated tool for designing, developing and realizing the electro- mechanical systems control. With more details we consider an hardware/software architecture which give us the capability of controlling external systems with analog inputs outputs (I/Os) and inputs for encoder/resolver signals. This tool shows like an integrated framework. It is interfaced with the Mat- Lab/Simulink tool which allows the users to control all the development process from the unique top-level interface. This tool introduces a powerful feature to the control development process: it is now possible to control the whole control design process from just one system. The capability for this tool of on-line interfacing control programs to Simulink, allows the user to directly check system and controller behaviors reducing in such a way the time needed for control adaptation.

An enterprise's ability to deliver new products quickly and efficiently to market is critical for competitive success. While manufactureres recognize the need for speed and flexibility to compete in this market place, companies do not have the time or capital to move to new automation technologies. The National Industrial Information Infrastructure Protocols Consortium's Solutions for MES Adaptable Replicable Technology (NIIIP SMART) subgroup is developing an information infrastructure to enable the integration and interoperation among Manufacturing Execution Systems (MES) and Enterprise Information Systems within an enterprise or among enterprises. The goal of these developments is an adaptable, affordable, reconfigurable, integratable manufacturing system. Key innovative aspects of NIIIP SMART are: (1) Design of an industry standard object model that represents the diverse aspects of MES. (2) Design of a distributed object network to support real-time information sharing. (3) Product data exchange based on STEP and EXPRESS (ISO 10303). (4) Application of workflow and knowledge management technologies to enact manufacturing and business procedures and policy. (5) Application of intelligent agents to support emergent factories. This paper illustrates how these technologies have been incorporated into the NIIIP SMART system architecture to enable the integration and interoperation of existing tools and future MES applications in a 'plug and play' environment.

Field device or process level networks are a major innovation in the field of controls. They are an application of digital communication networks to connect field devices such as actuators, sensors, and transducers with each other and other parts of control system like operator interfaces and controllers. The treatment of this subject has concentrated in the professional literature on the development and description of communication protocols as well as overviews of benefits they offer. It is written primarily by members of manufacturer and standards development communities for the end user. However, the topic is broader and includes issues related to implementation and end use experiences. This paper provides a convenient overview of field device networks and aspects of their impact on control system design.

The honing texture of gray iron cylinder bores of combustion engines mainly consists of two sets of approximately parallel, stochastically placed grooves at different angles to the cylinder axis. Oil consumption, longevity and noxious emissions highly depend on the quality of this groove texture, that is still often rated by visual inspection of microscopic images by an expert. This approach is tedious and subjective. Therefore it is desirable to develop methods which allow an objective, reproducible and automatic assessment of the honing texture. Within our paper we focus on one important aspect of the honing texture. A method for an objective quantification of the balance of the two groove sets is presented. The sets are compared with respect to their spatial groove densities, amplitudes, widths and angle deviation. A parametric, well- interpretable measure D is constructed that explicitly evaluates the discrepancy between both groove sets on the basis of their probability density functions regarding amplitudes, widths and locations. Additionally, a heuristical measure for the dispersion of the groove angles within each set is taken into account. We show the usefulness of the proposed approach by assessing synthetic textures and a series of real honing textures which are visually rated by experts. Small values of D are sufficient as well as necessary for a balanced appearance of a texture. Moreover, we analyze the statistical reliability of the measure with respect to the size of the observed texture area.

Global competition and rapidly changing customer requirements are forcing major changes in the production styles and configuration of manufacturing organizations. Agent-based systems are showing considerable potential as a new paradigm for agile manufacturing systems. With this approach, centralized and sequential manufacturing planning, scheduling, and control systems may be replaced by distributed intelligent systems to facilitate flexible and rapid response to changing production styles and variations in product requirements. In this paper, the characteristics and components of such a multi-agent architecture for advanced manufacturing are described. This architecture addresses agility in terms of the ability of the manufacturing system to solve manufacturing tasks using virtual enterprise mechanisms while maintaining concurrent information processing and control.

While current research into decentralized control architectures for advanced manufacturing systems has proven the concept of architectures ranging from hierarchical to heterarchical control, neither extreme of the control architecture spectrum has proven to be the most appropriate choice for given manufacturing systems. This paper reports on research into a control architecture that does show promise in this area: a hybrid control architecture consisting of elements of both hierarchical and heterarchical control architectures. First control architectural issues are considered. Next, the concepts of virtual control organizations and partial dynamic hierarchies are presented. The paper then concludes with a discussion of recent work on implementing partial dynamic hierarchies within real-time control constraints, using an agent-based architecture.

A holonic manufacturing system (HMS) is a manufacturing system where key elements, such as machines, cells, factories, parts, products, persons, teams, etc., are modeled as 'holons' having autonomous and cooperative properties. The distributed decision authority, the cooperative nature, and the integration of physical and informational aspects of system elements or holons make the HMS a new manufacturing paradigm, with great potential for meeting today's agile manufacturing challenges. Critical issues to be investigated include how to define holons for a given problem context, what should be the appropriate system architecture, and how to design effective cooperation mechanisms among holons for overall system performance. In this paper, holonic scheduling is developed for a factory consisting of multiple cells. Holons are identified, and their relationships are delineated through a novel modeling of the interactions among cells. The cooperation mechanisms among holons are established based on the 'Lagrangian relaxation technique' of mathematical optimization, and cooperation across cells is performed without accessing individual cells' local information nor intruding on their decision authority. The holonic system developed also possesses structural recursivity, and this property, combined with the integrability of individual holons, enables high system extendibility. Numerical testing shows that the method can generate high quality schedules in a timely fashion, and has comparable computational requirements as compared to a single-level Lagrangian relaxation method. The method thus provides a theoretical foundation for guiding the cooperation among holons, leading to globally near-optimal performance.

This paper describes three approaches to assigning tasks to resources. A fast and simple priority dispatch method is described and shown to produce acceptable schedules most of the time. A look ahead algorithm is then introduced that outperforms the dispatcher by about 12% with only a small increase in run time. These algorithms set the stage for the introduction of a genetic algorithm that uses job permutations as the population. The genetic approach presented here is novel in that it uses two additional binary variables, one to allow the dispatcher to occasionally skip a job in the queue and another to allow the dispatcher to occasionally allocate the worst legal position to the job. These variables are included in the recombination step in a natural way. The resulting schedules improve on the look ahead by as much as 15% at times and 3% on average. We define the 'window- constrained packing' problem and use it for simulations.