SHINE Expert System

Spacecraft Health Inference Engine (SHINE) is a software-development tool for knowledge-based systems and has been created as a product for research and development by the Artificial Intelligence Group, Information Systems Technology Section at NASA/JPL. The system is now in regular use in basic and applied AI research at JPL.

SHINE is a system developed at NASA to meet many of their demanding and rigorous AI goals for current and future needs. It is a system that was designed to be efficient enough to operate in a real-time environment and to be utilized by non-LISP applications written in conventional programming languages such as C and C++. These non-LISP applications can be running in a distributed computing environment on remote computers or on a computer that supports multiple programming languages.

SHINE provides a variety of facilities for the development of software modules for the primary functions in knowledge-based reasoning engines. The system may be used to develop artificial intelligence applications as well as specialized tools for research efforts.

Knowledge-based systems for automated task planning, monitoring, diagnosis and other applications require a variety of software modules based on artificial intelligence concepts and advanced programming techniques. The design and implementation of the modules require considerable programming talent and time and background in theoretical artificial intelligence. Sophisticated software development tools that can speed the research and development of new artificial intelligence applications are highly desirable. The SHINE system was developed for that purpose. Included in the system are facilities for developing reasoning processes, memory-data structures and knowledge bases, blackboard systems and spontaneous computation daemons.

SHINE is not a shell which means that the user does not have to solve his problems in the shell's way; you can solve them your own way. One is not limited by a shell's out dated problem solving techniques or preconceived notions on how things should be done.

Computational efficiency and high performance are especially critical in artificial intelligence software. That consideration has been an important objective for the SHINE system and has led to its design as a toolbox of AI facilities that may be used independently or collectively in the development of knowledge-based systems.

*SHINE comes with libraries that implement most common problem solving techniques and representations. This means that you can make use of classical AI solutions that have been extensively used and tested by other users. These libraries can also be extended by your own problem solving techniques and representations.

*SHINE facilities are invoked directly by a programmer in the Common LISP language. For improved efficiency, an optimizing compiler is included that generates highly optimized Common LISP code. SHINE allows embedded software written in other programming languages such as C, C++, and also permits software developed with the system to be part of larger, non-Common LISP applications.

*SHINE is a set of high level and low level software tools designed to assist in building stand-alone knowledge-based system applications, shells and tools.

*Since SHINE is a Common LISP based system, it can be run on any machine that supports Common LISP. Because SHINE is a compiler, not an interpreter, applications can run on a delivery machine that is much smaller than the development machine.

*SHINE provides the LISP programmer with the necessary tools to build a wide variety of reasoning and inference engines, such as expert systems, planners, diagnosticians, simulators, execution and real-time monitoring.

*SHINE allows and encourages embedded expert systems to be developed. This means that SHINE could be a supervisor of many other systems either written in SHINE or in conventional programming languages.

*SHINE is an optimizing compiler-based system. When an application is developed using SHINE, it is first translated into Common LISP code and then passed through an extensive optimizer. SHINE generates tailored code for each application. There are no intermediate levels of interpretation for execution unlike many commercial systems. SHINE programs are executed directly by the LISP interpreter and compiled directly by the LISP compiler. This means much greater speed and better portability to other machines.

*SHINE augments the Common LISP programming language and environment so that programs written in SHINE have direct use of all of the features of the underlying LISP system and the LISP environment.

*SHINE provides a debugging environment built upon the LISP environment for the debugging of systems at compile-time and run-time. In addition, the SHINE compiler and its run-time environment perform extensive error checking.

Background

Knowledge acquisition and implementation from experts is an inefficient and painful process for most automation implementation projects. The phase is often so wrought with difficulty, that the success of the automation project as a whole is jeopardized. The following describes a system called Spacecraft Health Inference Engine (SHINE) which provides a number of solutions to this problem.

SHINE is a state-of-the-art solution for Artificial Intelligence (AI) and non-AI problems that up to this point were either impossible or impractical to solve.

SHINE is a high-speed expert system and inference engine based upon the experience and requirements that were collected over the years by the Artificial Intelligence Research group at NASA/JPL in developing expert systems for the diagnosis of spacecraft health.

It is intended for those areas of inferencing where speed, portability and reuse is of critical importance. Such areas would include spacecraft monitoring, control and health, telecommunication analysis, medical analysis, financial and stock market analysis, fraud detection (e.g. banking and credit cards), robotics or basically any area where rapid and immediate response to high-speed and rapidly changing data is required.

SHINE was independently evaluated by UC Berkeley and was shown to significantly outperform commercially available inference engines such as RTI and ART. It executes approximately 500,000,000 plus rules a second running on a standard Windows PC computer.

*SHINE is written in Common LISP and can be easily run on any system that supports the language. It has been successfully interfaced to many non-LISP systems without any problems.

*VIASPACE [http://www.viaspace.com (www.viaspace.com)] ) has the Caltech licensing rights to most commercial applications of SHINE. They are currently working on both product and commercial enhancements to the SHINE technology as well as several Expert System applications in both the Defense and Security markets.

SHINE has been used in the following NASA and non-NASA applications:

1. Spacecraft Health Automatic Reasoning Pilot (SHARP) for the diagnosis of telecommunication anomalies during the Neptune Voyager (VGR) Encounter.

2. Galileo (GLL) mission for diagnosing problems in the Power and Pyro Subsystem (PPS).

3. Magellan (MGN) mission for diagnosis of telecommunication anomalies in the TELECOM subsystem.

4. Engineering Analysis Subsystem Environment (EASE) which is an operations environment to operate a large number of spacecraft simultaneously, maintain high reliability levels and increase productivity through shared resources and automation.

5. Extreme UltraViolet Explorer (EUVE) mission for labor 3 to 1 shift reductions through the use of artificial intelligence.

6. Fault Induced Document Officer (FIDO) for the EUVE mission. which is an automated system that assists in expert knowledge acquisition, access and publishing capabilities for safely managing complex systems under staffing reductions and "lights out" operations.

7. Stochastic Problem Obviation Tracker (SPOT) for the EUVE mission which captures and reports relevant statistical information to the user based on operations within the FIDO environment.

8. Under consideration by a medical company for real-time diagnosis of rectal colon cancer.

9. Under consideration by a medical company for an expert system for the control of the robotic systems used in eye surgery.

External links

* [http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/14039/1/00-0425.pdf SHINE use at NASA]
* SHINE has been commercially licensed to [http://www.viaspace.com VIASPACE Inc.] for commercial purposes