This paper develops a method for neural network control design with sliding modes in which robustness is inherent. Neural network control is formulated to become a class of variable structure (VSS) control. Sliding modes are used to determine best values for parameters in neural network learning rules, thereby robustness in learning control can be improved. A switching manifold is prescribed and the phase trajectory is demanded to satisfy both, the reaching condition and the sliding condition for sliding modes.

This paper presents a design methodology for parallel robots having tosatisfy a set of performance constraints. Some of these constraints are used tocompute a closed region in the parameters space (in which a point define anunique robot geometry) which define all the robot geometries fulfilling theseconstraints. Then a grid is created over this region and for each node of thisgrid the requirements, expressed in a high level language, are evaluated. Thenode leading to the robot best fulfilling the constraints is the designsolution.

A new real-time obstacle avoidance method for mobile robots has been developed. This method, namely the vector-distance function method, permits the detection of obstacles (both moving and stationary) and generates a path that can avoid collisions. The proposed approach expresses the distance information in a vector form. Then the notion of weighting is introduced to describe relationship between sensors of mobile robots and the target to be reached. Furthermore, R-mode, L-mode and T-mode are introduced to generate a safe path for the mobile robot in a dynamic environment filled with both stationary and moving obstacles. The algorithm can deal with a complicated obstacle environment, such as multiple concave and convex obstacles. Simulation results are included to demonstrate the applicability and effectiveness of the developed algorithm.

This paper presents a novel approach for seam tracking using ultrasonics. An ultrasonic seam tracking system has been developed for robotic welding which tracks a seam that curves freely on a two-dimensional surface. The seam is detected by scanning the area ahead of the torch and monitoring the amplitude of the waves received after reflection from the workpiece surface. Scanning is accomplished by using two ultrasonic sensors (a transmitter and a receiver) mounted on a stepper motor such that the transmitter angle is the same as the receiver angle. The motor is mounted on the end-effector just ahead of the welding torch and covers a ninety degree arc in front of the torch. If there is no seam then the receiver receives most of the transmitted waves after reflection, but if there is a seam then most of the transmitted waves are dispersed in directions other than that of the receiver. The system has been tested and is very robust in the harsh environments generated by the arc welding process. The robustness of the system stems from using various schemes such as time windowing, a waveguide, air and metal shields, and an intelligent sensor manager. This ultrasonic system offers some distinct advantages over traditional systems using vision and other sensing techniques. It can be used to weld very shiny surfaces, and is a very economical method in terms of cost as well as computational intensity. The system can be used to detect seams less than 0.5 mm wide and 0.5 mm deep.

This paper is concerned with a mobile robot reactive navigation in an unknown cluttered environment based on neural network and fuzzy logic. Reactive navigation is a mapping between sensory data and commands without planning. This article's task is to provide a steering command letting a mobile robot avoid a collision with obstacles. In this paper, the authors explain how to perform a currently perceptual space partitioning for a mobile robot by the use of an ART neural network, and then, how to build a 3-dimensional fuzzy controller for mobile robot reactive navigation. The results presented, whether experimented or simulation, show that our method is well adapted to this type of problem.

Conventional robotics has proved to be inflexible and non-generic. The concept of Distributed Manipulation Environment (DME) is introduced to overcome some of these shortcomings. This concept proposes a distributed approach to robotics and flexible automation. The work is concerned with modelling, simulation and event based control of DME. The modelling, conducted both at the atomic and the coupled level, is quite generic and provides a framework for static and dynamic behaviour analysis of DME systems. The simulation models serve as a mean of performance evaluation of the system on a computer before the actual implementation in real time. The event-based controller provides a simple and robust control scheme. The controller, itself, can be tested, validated and finely tuned through simulation before implementation. The feasibility of the modelling technique is demonstrated through a case study.

Based on the singularity-consistent parameterization framework, weanalyze motion at direct kinematics singularities for a broad class of parallelmanipulators. It will be shown that taking into account the instantaneous motiondirection of the output link, additional insight can be gained for thepossibility to move through such singularities. We argue that direct kinematicssingularities should be analyzed over the dual space which, in turn, involvesthe state of the passive joints. We perform such an analysis based on theconditioning of the equation of motion. It is shown that, depending on theinstantaneous motion direction, at certain direct kinematics singularities it ispossible to obtain a consistent solution in terms of torque. This implies thatin combination with the continuity of the singularity-consistent inversekinematic solution, motion through such direct kinematics singularities isfeasible.

This paper presents the design and development of a laboratory facility constituting a constrained planar flexible manipulator system. The system is designed for experimental investigations within research programmes involving flexible manipulator systems. An outline of a generic design procedure for flexible manipulator systems is given. The design criteria developed through this procedure account for the strength and stiffness of the manipulator and the required transducers and instrumentation. These are considered in detail and the system designed accordingly. Finally, practical problems encountered are highlighted and methods of dealing with such problems are presented and discussed.

This paper describes an interactive software system, developed at the Robotics and Automation Laboratory at UCLA to demonstrate innovative approaches to off-line robot programming and work-cell layout design. The software computes the time-optimal motions along specified paths, local optimal paths around an initial guess, and the global optimal path between given end-points. It considers the full robot dynamics, actuator constraints, on the payload acceleration or the gripping force, and any number of polygonal obstacles of any shape. The graphic displays provide a useful tool for interactive motion planning and workcell design.

The paper deals with the modelling of unknown objects and location determination of known ones. The operator models objects using data issued from the video image and a time of flight infra red range finder. We present methods to build an embodying volume for unknown objects. These methods are available for polyhedral and conical objects, which represent the basis of most industrial environments. The location determination of known objects is determined using 2D clues obtained through video image. A two-step algorithm is implemented and assessed. The first step is available in case of object large motion, it is a geometric algorithm, which doesn't use redundant data; it provides a first estimate of the pose. The second step is available in case of object small motion. A linear approach is carried out to determine the pose. It allows the use of redundant data, so it improves the accuracy achieved after the first step.

In this paper a model to cover all possible topologies of robot manipulators composed of prismatic and revolute joints is presented. For simplicity, only planar systems are considered, hence to provide plane positioning, systems handled are of three degrees of freedom. The physical model assumes three moving rigid links in articulation with one revolute and one prismatic joint between each link pair, forming a six degrees of freedom open chain linkage. Among each joint pair, one is real and the other fictitious. The real joint is arbitrarily actuated by an externally applied force or torque while the fictitious one is acted upon by an appropriately controlled force or torque as to keep that joint velocity zero, keeping fixed at its initial position. The physical model is accompanied by a mathematical model obtained by Lagrange formulation. This approach is called ‘The method of Fictititous Degrees of Freedom’.

The paper presents an object-oriented approach to the implementation of a software library (MRROC+) which contains building blocks for the construction of multi-robot system controllers tailored to meet specific demands of a task at hand. Moreover, the paper supplies a brief overview of robot programming methods.

For off-line programming to work, systematic methods must be developed to account for non-ideal performance of the parts and devices in the manufacturing cell. Although much of the literature focuses on robot inaccuracy, this paper considers practical methods for the tool control frame (TCF) calibration and rigid-body compensation required to close the inverse kinematics loop for target driven tasks.

In contrast to contemporary estimation methods, a closed-form, easily automated, solution is introduced for calibrating the position and orientation (pose) of orthogonal end-effectors when the distal robot joint is revolute. This paper also considers methods for measuring and compensating the small rigid-body perturbations that result from non-repeatable part delivery systems or from geometric distortion. These methods are designed to eliminate rθ error from the rigid-body prediction and can be conducted in real-time. Without accurate TCF calibration and rigid-body compensation, even the most accurate robot will fail to complete an off-line programmed task if the task tolerances are stringent.

This paper presents the kinematics and dynamics of asix-degree-of-freedom platform-type parallel manipulator with six revolute legs,i.e. each leg consists of two links that are connected by a revolute joint.Moreover, each leg is connected, in turn, to the base and moving platforms bymeans of universal and spherical joints, respectively. We first introduce akinematic model for the manipulator under study. Then, this model is used toderive the kinematics relations of the manipulator at the displacement, velocityand acceleration levels. Based on the proposed model, we develop the dynamicsequations of the manipulator using the method of the natural orthogonalcomplement. The implementation of the model is illustrated by computersimulation and numerical results are presented for a sample trajectory in theCartesian space.

1. Joint Venture Prograamme at Cranfield (UK)

The Joint Venture Programme, newly launched by the Cranfield School of Management, offers help to companies that are looking to expand their business in developing countries. It is claimed that valuable European Community (EU) grants are being lost, and the Cranfield initiative offers a fast track through what they describe as the bureaucracy involved in making applications for aid to develop partnerships abroad. This is particularly true of companies involved with automation and robotics who because of their intense development and marketing activities fail to take advantage not only of EU programmes that offer financial assistance, but even the help that is at hand from their own government's initiatives.

This paper addresses the tracking control problem of robotic manipulators with unknown and changing dynamics. In this study, nonlinear dynamics of the robotic manipulator is assumed to be unknown and a control scheme is developed to adaptively estimate the unknown manipulator dynamics utilizing generic artificial neural network models to approximate the underlying dynamics. Based on the error dynamics of the controller, a parameter update equation is derived for the adaptive ANN models and local stability properties of the controller are discussed. The proposed scheme is simulated and successfully tested for trajectory following tasks. The controller also demonstrates remarkable performance in adaptation to changes in manipulator dynamics.

In this paper, the development of a fuzzy logic control algorithm for walking machine leg control is presented. The planetary gear leg is designed for a quadruped which can walk and trot under the following design criteria: high efficiency, compact size and high payload/weight ratio. The inverse kinematics and dynamics of this leg are first analysed. A fuzzy logic control algorithm is then developed and implemented to control this prototype leg. The controller controls the leg using position control when the foot is in the air and force control when the foot is on the ground. This fuzzy logic control algorithm is evaluated in both simulation and experiments. A comparison of the proposed control algorithm with the conventional impedance control is given. Results show that the fuzzy logic controller is effective in controlling the leg machine.

A practical implementation of a genetic algorithm for routing a real autonomous robot through a changing environment is described. Moving around in a production plant the robot collects information about its environment and stores it in a temporal map, which is virtually a square grid, taking account of changing obstacles. The evolutional optimizer continuously searches for short paths in this map using string representations of paths as chromosomes. The main features of the implementation include physical realization, random walk exploration, temporal mapping, and dedicated genetic operators.

A calibration method for a Stewart platform has been developed as partof a project aimed at developing a calibration method for a Delta robot. TheDelta has 3 degrees of freedom (DOF) but is more complex than the Stewartplatform for calibration purposes because an extra link is inserted in eachkinematic chain between the base and the Nacelle member.

MDSF is a generic manipulator development and simulation facility developed by Spar Aerospace Limited, as a prime contractor to the Canadian Space Agency. It is an integral part of Canada's contribution to the International Space Station (ISS) program. MDSF is capable of simulating general robotic systems with complex and flexible structures undergoing all kinds of tasks including payload hand-over and contact with the environment. Rigorous validation and configuration control have been imposed throughout the development course of the simulation facility. Examples of current applications in the ISS program are included.