- Theory of hierarchical, distributed systems
The control of large systems is always organized in a distributed
hierarchy. The aim of this ONR-funded project is to develop a
theoretical framework for describing such control architectures, and
tools, techniques and applications to determine the utility of the
framework. The theoretical work will explore two innovations: a
`multi-world' semantics in contrast to the `one-world' semantics used
in control theory and AI today, and a semantics of `group of agents'
that is distinct from the `collection of agents'. In order to test the
utility of the framework we propose to develop tools for describing
such architectures based on the hybrid simulation language SHIFT, a
simulation testbed of distributed agents for the study of different
control organizations, and collaboration with other ONR researchers to
focus the tools on the control of unmanned vehicles.
Simulation-based optimization of Markov decision processes: An empirical
process theory approach.
R. Jain and P. Varaiya. Automatica 46 (2010) 1297-1304.
control of semi-autonomous teams under uncertainty. Mica Final Report v.5.
P. Varaiya et al. May 2004. [pdf 1.63M]
PAC learning for partially observable Markov decision processes.
R. Jain and P. Varaiya. Proc. Conf. on Information Sciences and Systems (CISS), March 2004. [pdf 111K]
under communication constraints.
R. Jain, T. Simsek and P. Varaiya. Proc. 41st IEEE CDC, Las Vegas,
Dec. 2002. [pdf 450K]
Communication and control of
distributed hybrid systems.
T. Simsek, J. Sousa and P. Varaiya. Proc. ACC, Washington, DC, 2001.
[pdf 260 K]
about hierarchical systems.
P. Varaiya. In T. Djaferis and I. Schick (eds), System Theory:
modeling, analysis and control, Kluwer, 2000. [pdf 54K]
A theory of hierarchical,
P. Varaiya and T. Simsek. Slides of talk presented at ONR review,
Los Angeles, CA. July 1998.
- Towards a layered view of control.
P. Varaiya. Slides of talk presented at 35th Alleron Conference, Urbana,
IL. October 1997. [postscript 437K].
The text can be found at