The traditional vehicle for the deployment of a real-time system has been a real-time operating system (RTOS). In recent years another programming approach has increasingly found its way into the real-time systems domain: the use of middleware. Examples are the so called pervasive systems (embedded, interactive but not mobile), and ubiquitous systems (embedded, interactive and mobile), e.g. hand-held devices. These tend to be dynamic systems that often exhibit a need for flexible scheduling because of their operating requirement; or their execution environment. Thus, today there is a true need in many realtime applications for more flexible scheduling than what is currently the stateof- prac'tice. By flexible scheduling we mean the ability of the program execution platform to provide a range of scheduling policies, all the way from hard real-time to soft real-time policies, from which an application can choose one most suited to its needs. Furthermore, some applications may need to be scheduled by one policy while others may need a different policy, e.g. fi'Ced priority or earliest deadline first (EDF) for hard real-time tasks, least slack time first (LST) or shortest remaining time for soft real-time tasks. It would be difficult for the middleware to expect this functionality from the RTOS. This would require a fine balance to be struck in the RTOS between flexibility and usability, and many years will probably pass until such approaches become mainstream and usable. 'This thesis maintains that this flexibility can be introduced into the middleware. It presents a viable solution to introducing flexible scheduling in real-time program execution middleware in the form of a flexible scheduling framework. Such a framework allows use of the same program execution middleware for a variety of applications - soft, firm and hard. In particular, the framework allows different scheduling policies to co-exist in the system and their tasks to share common resources. The thesis describes tlle framework's protocol, examines the different types of scheduling policies that can be supported, tests its correctness through the use of a model checker and evaluates the proposed framework by measuring its execution cost overhead. The framework is deemed appropriate for the types of real-time applications that need the services of flexible scheduling.

Virtual textile databases provide a cost-efficient alternative to the use of existing hardcover sample catalogues. By taking advantage of the high performance features offered by the latest generation of programmable graphics accelerator boards, it is possible to combine photometric stereo methods with 3D visualisation methods to implement a virtual textile database. In this thesis, we investigate and combine rotation invariant texture retrieval with interactive visualisation techniques. We use a 3D surface representation that is a generic data representation that allows us to combine real-time interactive 3D visualisation methods with present day texture retrieval methods. We begin by investigating the most suitable data format for the 3D surface representation and identify relief-mapping combined with Bézier surfaces as the most suitable 3D surface representations for our needs, and go on to describe how these representation can be combined for real-time rendering. We then investigate ten different methods of implementing rotation invariant texture retrieval using feature vectors. These results show that first order statistics in the form of histogram data are very effective for discriminating colour albedo information, while rotation invariant gradient maps are effective for distinguishing between different types of micro-geometry using either first or second order statistics.

No description available.

No description available.

No description available.

No description available.

No description available.

Variable structure control is a kind of nonlinear control with discontinuous input. The control input switches from one value to another according to the system states. Its usefulness of robotic control is the existence of the switching surface, since once the system states are sliding on the switching surface, the control is insensitive to the parameter uncertainty and external disturbances. The application of variable structure control for robot manipulators has experienced several stages. Most efforts have been to make this control method practically more applicable to robotics control problems. In this thesis, variable structure control is studied again, but from another angle. The computational efficiency when this control method is applied to robots in real-time is considered. A Newton Euler formation of a model based variable structure controller is developed in this thesis. The control is based on the model based computed torque control and a variable structure control component is added into it to cope with parametric uncertainty. The control can be written in the form of the addition of two Newton Euler formulations of robot dynamic equations. In the form of the Newton Euler formulation the control is more computationally efficient and the structure of the control as the addition of two Newton Euler formulations makes it possible to compute the control in parallel. The control formula is further extended to Cartesian space in the hope that it can be used in conjunction with external Cartesian space sensors. The control is studied firstly by simulations which provide the initial indication of its usefulness, then further demonstrated on the first three joints of an electrical driven PUMA 560 arm. In the real time implementation of this control, the control performance is compared with purely model based computed torque control under several conditions. Both simulation and real time implementation results are presented. In the real time implementation, further computational reduction of the control is achieved by considering the characteristics of the control and the Newton Euler formulation of the robot dynamic equation.

No description available.

No description available.