The recent developments of communication standards such as MAP, along with progress in CAD/CAM and CIM have introduced new requirements and applications for Computer Numerical Control. The current Computer Numerical Controller (CNC) is the result of an evolution spreading over the past 30 years and we feel that it no longer has an adequate structure to meet its current requirements. In this paper, we present an analysis of the problems and limitations we have found with today's CNCs. We then present revolutionary (as opposed to evolutionary) concepts for their structure. Ease of configuration, higher level programming, integration in the manufacturing cell, possibility of improving tooling feedback combined with a knowledge-based architecture are the driving forces behind our design.

Binary decision (BD) programs have widespread applicability in diverse areas. In many situations, problems are defined in terms of Boolean expressions, which must be converted to decision programs for evaluation. Since the construction of optimal BD programs is NP-complete, absolute optimization appears computationally intractable. This paper presents a fast heuristic algorithm for constructing near-optimal decision programs and provides an optimality metric. The algorithm involves two steps: preprocessing and optimization. The preprocessor builds a sub-optimal (in some conditions near-optimal) decision program in linear time. If sub-optimal programs are generated, the optimizer is invoked, producing a near-optimal program, using decision tables, in 0(n2) time, where n is the size of the reduced decision table generated in the first step.

This paper shows the development of a control strategy for a digitally controlled Michelson Interferometer inside of a MIDAC Fourier Transform Infrared (FTIR) spectrometer. Initial investigation has shown undesirable response characteristic, primarily due to locations of the zeros of the system. The effect of the zeros on a general second order case will be studied and the definitions of the poles and zeros for the multivariable system are summarized. Techniques to compensate for the undesired location of zeros are derived and applied to the control of the interferometer under investigation.

Recently various dynamic control algorithms for industrial manipulators have been proposed. However, computation time, modeling error, and torque type servo controller design prevented real-time implementation. As a result, most of performance evaluations of the dynamic control algorithms were carried out only by computer simulations. In this paper, we explore real-time implementation of dynamic control algorithms for industrial manipulators to show the feasibility and effectiveness of such algorithms. Experimental results indicate that computed torque technique and iterative learning control algorithms can be effectively applied to controlling industrial manipulators.

A simple and direct method to compensate unknown load effects on manipulator motions by a six-axis force sensor installed between end-effector and the load is proposed for trajectory control of robot manipulators. This method can also compensate any external disturbance forces and moments imposed on end-effector. The validity of the method greatly depends upon the performances of the force sensor. Use being made of a recently commercially available six-axis force sensor, experiments of trajectory control for PUMA type manipulator are examined. The results show that the six-axis force sensor works well to compensate the unknown load effects and the method is useful for trajectory control of the manipulator.

In this paper the synthesis of a decentralized control for a robotic system is performed. In this control the robotic system is considered like a set of independent subsystems each corresponding to a separate joint and coupling among them is neglected. We have developed a computer program for obtaining a local controller that stabilizes the local subsystem. This local controller is based in the pole placement dynamic controller. For this, we have obtained the mathematical model of each actuator of robot in the state domain, taking into account the inertia forces of the mechanical part around the i-th joint.

An extension of Data Flow Diagram as appied to system software analysis is proposed for the analysis and representation of microcomputer based automatic process control systems. The proposed Data and Control Flow Diagram is applied for the analysis and representation of a Vertically Hierarchical Distributed Process Control System to demonstrate the manner in which it facilitates the analysis, implementation and representation of such systems.

Distributed systems have been proposed for the simulator design. However, to get most benefit from these systems both hardware and software have to be carefully designed. In this paper the system architecture and some developed parallel integration algorithms are presented and compared. Real-time implementability, stability, and speed are the evaluation criteria. Future research in the area is discussed.

This paper contains a discussion of numerical integration methods with regression coefficients that are matrices rather than the usual scalars. Methods of this type will be shown to be particularly useful for real-time simulations as they allow arbitrary placement of the simulation closed loop poles. This property is generally very useful, but especially so for the real-time simulation of stiff systems. Linear systems will be considered initially, followed by obvious extensions to nonlinear systems. Adaptive integration using these methods will also be introduced.

This paper describes a method for modelling industrial robots that considers dynamic approach to manipulation systems motion generation, obtaining the complete dynamic model for the mechanic part of the robot and taking into account the dynamic effect of actuators acting at the joints. For a four degree of freedom SCARA robot we obtain the dynamic model for the basic (minimal) configuration, that is, the three degrees of freedom that allow us to place the robot end effector in a desired point, using the Lagrange Method to obtain the dynamic equations in matrix form. The manipulator is considered to be a set of rigid bodies inter-connected by joints in the form of simple kinematic pairs. Then, the state space model is obtained for the actuators that move the robot joints, uniting the models of the single actuators, that is, two DC permanent magnet servomotors and an electrohydraulic actuator. Finally, using a computer simulation program written in FORTRAN language, we can compute the matrices of the complete model.

This paper presents the motion of a robot when an adaptive control strategy is used for position control system. This technique assures the accuracy of motion control system for a given input reference. To compensate the effects of nonlinearity and noise in the system, an adaptive procedure is proposed. The change in the inertia can be formulated as a nonlinearity factor when a model is considered for the system. This control strategy can be used for all the axis control of the robot when a particular trajectory is followed. The results of simulation shows that an accurate motion control system is obtained.

This paper presents the design of an autonomous navigation system (NAUTILUS) and its simulation. The NAUTILUS system consists of four parts: (i) a triangle base with three program-mable wheels, (ii) a special digital compass based on the Grey code, (iii) a sonar system in a hexagonal configuration, and (iv) a microprocessor-controller which selects information from the above components and determines the navigation paths.

An implicit high-level control language was designed and implemented for describing and controlling the path to be executed by the FMC Autonomous Vehicle. This language is an interpreted command language and runs on a Sun workstation. It takes a list of points describing the robot's global path in either Universal Tansverse Mercator (UTM) or geodetic (latitude, longitude) coordinates. First, the input is interpreted into descriptions of vehicle direction, a sequence of bounded path segments, and advice for road following and other such actions. Then, local sensor data is fused with this global path description and finally translated into mobility commands for vehicle control.

The new pragmatic method for designing stable model reference adaptive system (MRAS) is given in this paper. As the lemma, the sufficient condition assuring the asymptotic hyper-stability in the whole MRAS and its mathematical proof are presented. An example, computer simulation and case study show that this method is correct and reasonable. By employing this method, the difficulties of modifying adjustable parameters can be avoided, Erzberger's condition may not be taken into consideration and the structure produced by this method is simpler than that produced by the adaptive model reference following control system (AMFC). It is a feasible scheme for practical engineering.

An adaptive controller has been designed to control the pressure and force of a testing process of the glass viscoelastic property. The testing process appears to be nonlinear with the characteristic parameters varying drastically from test to test due to the difference of the composition of the testing glass samples and the variation of testing conditions. The proposed control scheme includes a simplified nonlinear model with unknown parameters, an on-line parameter estimator and an adaptive control algorithm combining feedforward prediction and PI (proportional plus integral) controller. Experimental results prove that the designed adaptive controller is adequate to control the glass testing process over a very wide range of testing conditions.

Conventional adaptive autopilots adapt themselves to the changes of the ship dynamics and keep the heading in the desired direction even if the ship dynamics changes. The paper describes an adaptive autopilot with new algorithms which estimate the disturbances caused by the water waves and feedforward the estimated values to the controller with the feedforward gain adapted to the changes of the water wave characteristics. The algorithms have been proved to be effective by computer simulations and on-board tests.

A new adaptive scheme is presented to estimate the state vector and identify the variable parameters of a single-input, single-output, nth order linear system; The approach taken here proceeds from a discrete formulation of the problem and a complete separation between the identification and estimation steps. One originality of this approach is the on-line determination of the observer parameters. Conversely to the other methods the observer design becomes very easy and can be extended to multivariable systems.

This paper presents a new approach for the Adaptive Control, having affect on Identification of system under control, holding at the block control the idea given by Clarke and Gawthrop, the Self Tuning Controler. From this, for the Identification block, the least squares method or its recursive version, is replaced by the identification ones known as least squares LATTICE (or LADDER in accordance with other authors) or its recursive version.

The control of robot manipulators is a very complex problem. The paper presents adaptive control method and robust adaptive control method with feedforward compensation for robot manipulators. The proposed two methods ensure the system asymptotically stable and have the advantages of strong anti-interferents and quick tracking to the desired trajectary compared with other methods.

This paper addresses the problem of controlling a flexible limb robot with an unknown load. The robot considered contains a single link. A lumped-parameter model is developed and a scheme is presented for rapid identification of these parameters (including load). Algorithms are then presented for updating the state estimator and the controller as the load changes. A complete adaptive closed-loop system is proposed.

A variable structure control scheme which consists of continuous adaptive gain feedback and feedforward controls with an equivalent external diturbance observer is developed to achieve accurate (robust) decoupled model following control for a two-degrees-of-freedom manipulator powered by PWM transistor converter-fed dc servomotors. The proposed control scheme is implemented by a fast digital signal processor(DSP). It is confirmed by the experiments that the position trajectories are smooth and track the desired trajectories accurately(robustly).

This paper proposes two pole assignment control schemes for robotic manipulators, based on an anticipatory action. In one, the control objective is for the velocity tracking error to decay with a prespecified dynamics. In the other, a generalised cost function is minimized and the weighting factors in the cost function are determined to achieve desired closed loop pole locations for the tracking error. The prediction scheme used ensures a high degree of robustness against system-model mismatch as demonstrated by the simulation results presented.

Electrical machines are supplied with electric power, therefore some modulation of electric power is required for control of them. This modulation is mainly achieved by PWM (Pulse Width Modulation), in which continuous signal is represented by the sequence of "ON" and "OFF" signal. However, any attention has not been paid to this modulation from the control point of view. The control systems of electrical machines should be optimized including the power modulation parts. VSS (Variable Structure Systems) are suited for such systems as require some modulation of electric power, because they are essentially based on discontinuous control. Motion control systems using electrical machines are designed to be robust by VSS taking discontinuous signals of VSS as "ON" and "OFF" of switching devices. In this paper, validities of VSS applying to electrical machines are discussed based on experiments of actual plants.

A strategy for position control of a high torque brushless dc drive, based on the variable structure method is proposed. This controller retains variable structure switching in the large, but near the equilibrium point, introduces smoothing to remove chattering and integral action for offset rejection. The method has been implemented experimentally and the step response under various loading conditions were compared with results obtained from a conventional variable structure design and a state space design.

A new variable structure control scheme which consists of continuous adaptive gain feedback(PI) and feedforward controls is developed to achieve accurate decoupled model following in a class of nonlinear time-varying systems in the presence of disturbances, parameter variations and nonlinear dynamic interactions. Then the developed method is practically applied to decoupled model following motion control for a two-degrees-of-freedom manipulator powered by PWM transistor converter-fed servo motors. The overall control strategies are implemented with a DSP(TI,TMS32020). Tt is found by the experiments that this controller is simple, is easily designed and performs extremely satisfactorily.