Intelligent Systems: Safety, Reliability and Maintainability Issues

This book is a collection of some of the papers that were presented during a NATO Advanced Research Workshop (ARW) on "Intelligent Systems: Safety, Reliability and Maintainability Issues" that was held in Kusadasi, Turkey during August 24- 28, 1992. Attendance at this workshop was mainly by invitation only, drawing people internationally representing industry, government and the academic community. Many of the participants were internationally recognized leaders in the topic of the workshop. The purpose of the ARW was to bring together a highly distinguished group of people with the express purpose of debating where the issues of safety, reliability and maintainability place direct and tangible constraints on the development of intelligent systems. As a consequence, one of the major debating points in the ARW was the definition of intelligence, intelligent behaviour and their relation to complex dynamic systems. Two major conclusions evolved from the ARW are: 1. A continued need exists to develop formal, theoretical frameworks for the architecture of such systems, together with a reflection on the concept of intelligence. 2. There is a need to focus greater attention to the role that the human play in controlling intelligent systems. The workshop began by considering the typical features of an intelligent system. The complexity associated with multi-resolutional architectures was then discussed, leading to the identification of a necessity for the use of a combinatorial synthesis/approach. This was followed by a session on human interface issues.

This book contains papers presented at a NATO ARW on issues relating to intelligent systems. Topics addressed include analysis and assessment issues, dependability, intelligent front ends, learning systems, safety and reliability of advanced robotic systems, and dependability considerations in industrial applications.

Reliability of Intelligence: Issues and Conjectures.- Multiresolutional System: Complexity and Reliability.- MARSA - the Software Tool for Reliablity and Safety Assessment of Control Systems.- Safety Aspects of Security Models.- The Intelligent Use of Intelligent Systems.- Intelligent Systems: The Weakest Link?.- Developing Hypermedia Front-Ends for Robot Teleoperation Tasks.- Automated Inspection and Reverse Engineering.- Development of the Intelligent Graphical Front-End for Robot Grasp Planning.- Another FormalApproach to Modelling Inference Engine Behaviour.- Fractal Set Theoretic Analysis of Performance Losses for Tuning Training Data in Learning Systems.- Automatic Construction of Qualitative Models.- Singleand Multiple Memory-based Neural Networks for Efficient Function Approximation.- Robustness and Fault Tolerance Issues in Artificial Neural Network Based Robotic Control Applications.- Human Factors in Robotic Systems.- Advanced Robotics - Towards Intelligent Machines.- Intelligent Motion Planning for Safe and Reliable Control of Robot Arms.- Control Issues of the Delft Intelligent Assembly Cell with Emphasis on Geometrical Trajectory Planning.- Deep Knowledge and Rigorous Engineering Practice: Two Ingredients to Build Dependable Knowledge Based Systems for Industrial Process Plant Control.- Architectural Foundations, Concepts, and Methods Behind ISACS - Real-Time Intelligent System for Critical Applications.- Intelligent Industrial Control.- Modelling of the Process of Air Traffic Control Underthe Expert System Management.