Next: Stochastic Adaptive Control Up: SYSTEMS AND CONTROL Previous: Adaptive Identification in

COCOLOG: Foundations, Theory and Implementation

Authors: [tex2html_wrap4412]P.E. Caines, D. Dyck, T. Mackling, C. Mascarua, Y-J. Wei

Investigator username: peterc

Category: systems and control theory


The acronym COCOLOG stands for (families of first order) conditional observer and controller logical theories. In the foundational work with S. Wang, problems of observation and control for partially observed input-state-output machines were formulated in terms of COCOLOG trees of first order logical theories. Subsequent work with Y-J. Wei has resulted in the definition of the so-called Markovian fragment systems of full COCOLOG theories; these have the property that they have virtually the same control decision making power as full COCOLOG theories but their axiom sets are of constant size over time. This is important for automatic theorem proving (ATP) implementations.

Current work is focusing on hierarchies of COCOLOG control theories which express and perform the control of complex systems via their formulation as hierarchical control lattices. These notions form a foundation for work with C. Mascarua in which COCOLOG is being extended to give expression and solution to certain geometric control problems such as those arising in robotics.

Implementation work has involved the creation by T. Mackling of the Blitzensturm series of efficient ATP programs; these are being applied to logic control problems in work with D. Dyck, T. Mackling, and Y-J. Wei.