Knowledge-Based Systems for Computational Control
Abstract
The goal of the project has been to create new tools for control system design and to establish 'pathfinders' for implementing new design principles in operational control systems. Four topics were addressed in the subject project: Neural Networks for System Modeling and Nonlinear Control; Stochastic Robustness of Control Systems; Computer-Aided Control System Design; Optimal Rule-Based Guidance for Autonomous Vehicles. The principal result of the first task is the development of a new method for training neural networks using extended Kalman filtering to match not only multivariate functions but the gradients of their surfaces. The principal result of the second task is the development of a powerful new method for characterizing the robustness of control systems and for designing controllers with satisfactory stability and performance robustness. The principal result of the third task is the identification of a new computational structure for multidisciplinary control system design. The principal result of the fourth task is a new approach for designing real-time rule-based controllers that can operate in uncertain environments, with particular application to the guidance of vehicles on a roadway with traffic.... Neural networks, Tobust control, Computer-aided control system design, Autonomous vehicles, Intelligent control.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1992
- Accession Number
- ADA260409
Entities
People
- Robert F. Stengel
Organizations
- Princeton University