RALPH-MEA
Architecture
Philosophy and Methodological Assumptions
Description of the Multiple-Execution Architecture
Decision Theoretic Control
Goal-based
Action Utility
Condition Action
Mulitple Knowledge Types
Meta-Level Arbiter
Executor/Monitor
Replanning Component
Agent Properties
Style of Control (Utility Driven)
Parallel Execution
Serial Processing
Modular Organization
Hierarchical Organization
Symbolic World Model
Black Box Approach
Declarative Representations
Global Representation and Uniform Access to Knowledge
Heterogeneous Knowledge Representation
Flat Planning (no Means-End Analysis Technique)
Agent Behavior
Coherent Behavior
Reflexive Response to Stimuli (Reactive Response)
Deliberation/Operation Speed
List of Agent Properties for All Architectures
Capabilities
Planning and Situated Action
Replanning
Meta-Reasoning
Prediction (Environmentally Constrained Reasoning)
Real-Time Execution
Support of Multiple, Simultaneous Goals
Focused Behavior and Processing/Selective Attention
Responding Intelligently to Interrupts and Failures
Support for Inaccurate Sensing
Robotic Tasks
List of Capabilities for All Architectures
Environmental Considerations
Dynamic Environments
Simulated Environments
Unpredictable
Varying Priority
Limited Response Time
Markov Assumption
Supervisors (provision for Utility Functions)
List of Environmental Considerations for All Architectures
Issues
References
Table of Contents
Current Location: RALPH-MEA - Top