Annual Review in Automatic Programming最新文献

In this paper, we present a new control system for the intelligent force control of multifingered robot grips which combines both fuzzy-based adaptation level and a neural-based one with a conventional PID-controller. The most attention is given to the neural-based force adaptation level implemented by three-layered back-propagation neural networks. A computer based simulation system for the peg-in-hole insertion task is developed to analyze the capabilities of the neural controllers. Their behaviour is discussed by comparing them to conventional and fuzzy-based force controllers performing the same task.

In a distributed robot system, asynchronous and synchronous communication between the system components is necessary to guarantee problem soliving capability in realtime. On that account, the distributed control architecure of the Karlsruhe Autonomous Mobile Robot KAMRO which is being developed at IPR has been extended by these communication kinds. The robot system consists of several subcomponents, like manipulators, two hand-eyecameras, one overhead camera and a mobile platform. To get better problem solving capability than the former centralized control architecture, these components are able to work together in teams (asynchronous communication) or special agents (synchronous communication).

Our research deals with the intelligent supervision and control of batch fermentation processes. We built a pilot plant, with all the auxiliar electronics needed, for a MS-DOS computer to interact with the plant. We selected KAPPA-PC as the main platform to develop a real-time application. It is an object-oriented development environment, with the tools to define and manage class hierarchies including sets of rules, for intelligent functions. The shell works under MS-Windows, and is open to userdeveloped specific C code. After some experimental modelling, we determined models, and defined intelligent classes to simulate the plant, to control the process, and to supervise it, in real-time.

Data Flow is a natural paradigm representing distributed programs able to express the data (and control) dependencies between composing entities (nodes). Our work concentrates on Data Flow programs having Hard-Real-Time constraints; we consider periodic programs but the method can also be applied to sporadic and aperiodic ones. We propose a method to find, if possible, a feasible execution time assignment for each program node in the system, given a heuristic mapping.

To compare realtime operating systems one must often rely on specifications given by the manufacturers. These published data are often measured by so-called benchmarks. But with the help of benchmarks it is not possible to collect sufficient information to determine the system behaviour for worst cases. Therefore it must be clarified which latency times exist and how they arise. Of great significance is the additional latency time of the operating system, which is compulsory for the design of the application. In this paper it is investigated whether and how long a given task, activated by an external interrupt, is delayed until task completion. For that all latency times which can occur are discussed, specifically the worst cases. We are taking a closer look at realtime operating systems which encourage so-called kernel-level threads. The advantage of these kernel-level threads systems in contrast to traditional systems are discussed referring to performance. At the moment we are testing two systems, LynxOS and HP-RT. We want to measure the worst-case latency times and therefore illustrate how this is done.

Ruled based systems have proven to be suitable for monitoring and diagnosis in process industry. However, up to now, they do not take into account the effect of their recommendations on the process under study, which is essential for achieving a total automatization of an idustrial plant. Besides, cooperation with other AI techniques within a consistent framework seems essential for tackling all aspects of industry problem. The system described in this paper tries to adress these two issues. The system has been installed in a petrochemical plant in Cartagena, Spain, showing very promising results.

Current verification tools for critical systems require the test engineers to be fully acquainted with several areas such as the physical systems for which the tests are being defined, and programming techniques, since engineering concepts are very far from computing. The implementation of front-end tools enabled to accept tests described in Natural Language and graphics may help to reduce this distance.

Recently increasingly artificial intelligent methods are used in the computer integrated manufacturing (CIM) applications. Real-time expert system shells are good tools to develop intelligent cell controllers. The first part of this paper summarises the communication problems of the expert systems in the CIM applications in the level of connection point of view. Then a research application called SSQA are shortly described with its aims and modules. The third part explains the way of the communication in SSQA.

A new and powerful approach to threading is proposed, that is designed to improve the responsiveness of concurrent logic programs for distributed, real-time AI applications. The technique builds on previously proposed scheduling techniques to improve responsiveness by synchronously passing control and data directly from a producer to a consumer. Furthermore, synchronous transfer of data requires less buffering and so less garbage is produced. Arguments are also passed in registers, further reducing overheads.

Complex systems are large applications, typically running on distributed, heterogeneous networks, driven by a number of distinct constraints and desiderata on goals such as performance, real-time behavior, and fault tolerance. These requirements frequently conflict, and satisfaction of these design objectives interacts strongly with assignment of system tasks to processors. The NSWC design framework DESTINATION provides an assignment module which can be used to optimize the system, as measured by the value of a weighted combination of objective cost functions.

For even modest-sized systems and networks, assignment space is too large to search exhaustively. We have implemented algorithms which generate heuristically good assignments. However, compile-time evaluation of many interesting design factors, even those clearly related to assignment, is impossible without some estimate of the schedule. We therefore provide approaches for determining a reasonable “pseudo-schedule” for a given system, network, and assignment, and use this to simulate execution in evaluating cost functions.