Download Artificial Intelligence in Real-Time Control 1991 by M. G. Rodd, G. J. Suski PDF

By M. G. Rodd, G. J. Suski

This set of court cases comprises the main major papers awarded on the 3rd IFAC Workshop on man made Intelligence in Real-time keep watch over, which used to be held from September 23-25, 1991 within the united states. during this workshop, even if there have been nonetheless a few "exotic" purposes, a more effective view of the functions and obstacles of present AI know-how ruled the individuals' discussions. With its resultant specialize in reliability and safeguard concerns, the workshop posed as many questions because it spoke back. It presents an outstanding replicate of the present cutting-edge which those complaints are meant to demonstrate

Show description

Read Online or Download Artificial Intelligence in Real-Time Control 1991 PDF

Similar nonfiction_12 books

Surface Modification Technologies XIII: Proceedings of a conference held in Singapore, 7-10 October 1999

The forty six papers provided at this occasion disguise tough fabrics, Lasers and floor Melting, Electrodeposition and review of converted Surfaces, Thermal Spray options, Nitride Coatings and overview of converted Surfaces.

Additional info for Artificial Intelligence in Real-Time Control 1991

Sample text

This is achieved by running simulation with a highly abstracted model, and then zooming into faulty abstract components, and continuing down in the hierarchy of models. By doing that, it implements a very flexible method to refine the knowledge about the state of the system in accordance with the current deadlines. To implement our work we used the BB1 blackboard shell, a general purpose framework for building intelligent systems. This tool incorporates a rich representational language to build models and a very clean and powerful architecture to program and execute monitoring, reasoning and action operations.

After completing one component in the domain model, the algorithm proceeds to the next one in the direction of the process if the algorithm is executed forward (forward means that causes, links are used to propagate causal links), taking as inputs the output signs of the preceding component, and repeating the same process using the prime model of the new component. A similar procedure is executed to implement backwards model processing, using the caused-by links instead of causes links, and moving in the sequence of structures and processes backwards.

The long-term goal of this project is to explore the utility of these ideas in controlling human posture. Keywords: Artificial Neural Networks, Adaptive Critic, Functional Link, Reinforcement Learning, Cart-Pole system. involving the balance and stability of flexible structures, such as space-borne structures, twolegged walking robots, aiming of rocket thrusters, and so on. The power of classical adaptive control techniques to solve these problems is limited because they work best when the system parameters are known and when the dynamic equations are linear and deterministic.

Download PDF sample

Rated 4.84 of 5 – based on 14 votes