Constraint-based computational intelligence

点击翻译按钮获取中文摘要

This research program is focused on constraint-based computational intelligence. We have developed a formal theory for the design and implementation of constraint-based intelligent systems. Our Constraint-Based agent (CBA) framework is a set of theoretical and practical tools for designing, simulating, building, verifying, optimizing, learning and debugging controllers for agents embedded in an active environment. The agent and the environment are modelled symmetrically as, possibly hybrid, systems in the Constraint Net (CN) dynamical systems language, as developed in this research program. We intend to develop and apply the CBA framework, emphasizing the important special case where the agent is an online constraint-satisfying device. Here it is often possible to verify complex agents as obeying real-time temporal constraint specifications and, sometimes, to synthesize controllers automatically. The CBA framework demonstrates the power of viewing constraint programming as the creation of online constraint-solvers for dynamic constraints.Our objectives over the next five years are to develop the theory further and to validate and improve it using experimental implementations on varous robotic systems in our lab. This research program is focused on constraint-based computational intelligence. We have developed a formal theory for the design and implementation of constraint-based intelligent systems.The experimental environment for this work includes the soccer-playing platforms pioneered by us, a CBA Java environment and a set of mobile robots with on-board real-time stereo and colour vision. We propose to use CBA models for the control, perception and coordination algorithms required for multi-agent real-time collaborative activities such as soccer, navigation and manipulation. The scientific novelty of this approach stems from its formal foundation, its practicality and its simplicity as a unitary theory of hybrid intelligent systems. The symmetry of the agent-environment pair model is necessary and unique. The technical significance is that it is a foundation for a new design framework for perceptual, collaborative, constraint-based agents.