Advanced real-time embedded system tools for intelligent architectures (ARTESIA)

Abstract

Existing knowledge-based systems and development tools were never designed for real-time embedded computing environments since they have historically been used in non-real-time applications. Furthermore, fundamental design concepts for real-time performance have been virtually ignored in the design of languages, shells, and tools. Real-time knowledge-based systems pose unique engineering problems which must be identified in order to resolve these shortcomings. Our research has focused on developing a set of tools to support three real-time AI processing objectives: (1) real-time intelligent control of system resources; (2) predictable real-time knowledge-based system performance; and (3) demonstration of these technologies in an embedded environment running a real-time operating system. Reticular Systems, Inc. Has developed a set of software tools for building real-time intelligent systems. This toolset is called ARTESIA: Advanced Real-Time Embedded System Tools for Intelligent Architectures. These tools include a low level real-time executive for intelligent control called TaskMaster/sup TM/. TaskMaster/sup TM/ works with the real-time operating system (RTOS) and the knowledge-based programs that run on the RTOS to ensure that the most critical application tasks are accomplished prior to their respective deadlines and that all tasks are completed in a timely manner. TaskMaster/sup TM/ accomplishes intelligent control by utilizing high-level knowledge about the application program's goals and beliefs as well as the relative priority of the applications. This information is used to generate schedules.