Optimization has long been used as a basis for control. Independently, constraint programming has been applied to many domains such as configuration design, planning, and scheduling. However, theories and algorithms for constraint programming and control have been developed in separate communities and have emphasized different problem areas. Constraint programming research has focused more on languages and constraint satisfaction techniques in discrete domains, while control research has concentrated more on numerical solutions to linear and non-linear constrained optimization problems. With the dramatic increase of computation power and the decrease of chip sizes and cost, intelligent embedded control becomes more an integral and center part of smart devices. The purpose of this workshop is to bring researchers from constraint programming and control together in order to create a synergy between the two communities, to explore new methods of programming, optimizing, reasoning, verifying and diagnosing of embedded control systems with both computational and physical constraints.

The workshop consists of five scheduled presentations and three invited talks on October 16, 1999. Three invited talks are given by

• Prof. Anil Nerode from Cornell University,
• Prof. Sunil Agrawal from University of Delaware,
• Ms. Elena Messina  from NIST

The workshop is intended to be an informal gathering of researchers who work in the area of constraint programming and constraint-based control to exchange information on current problems, existing solutions, and future research topics. 

Schedule and Papers

Abstracts of Invited Talks:

Title:  Real-Time Control Systems Architectures

Abstract:

We have been investigating means of more formally representing and analyzing complex, real-time architectures, such as the Real-Time Control System (RCS) Architecture. A category of formal and semi-formal languages and tools, called Architectural Description Languages (ADLs) is emerging.  We have been studying the applicability of ADLs for the representation, analysis, and validation of RCS.  We created a prototype using an ADL called Rapide, from Stanford University. The prototype defined a node of the RCS architecture as applied to autonomous scout vehicles.   The focus of our study was not constraints-based per se, but Rapide does have provisions for defining constraints and simulating the behavior of an architecture. This talk will present some possibilities in the area of tools that help developers of control systems that include constraints.   We will also briefly describe RCS and show a videotape of its application to military scout vehicles.

Brief Bio:

Elena Messina is currently Group Leader of the Knowledge Systems Group in the Intelligent Systems Division at the National Institute of Standards and Technology (NIST).   Her group's work includes knowledge representation and planning for control as well as  software engineering, tools, and methodologies for building controllers. Current development testbeds include an autonomous scout vehicle and an inspection workstation with feature-based planning, control, and part localization.  Prior to joining NIST, she was Manager of Geometric Modeling for the I-DEAS Master Series CAD/CAE Software at the Structural Dynamics Research Corporation.   Ms. Messina also worked at Cincinnati Milacron on their industrial robots, where she received 2 patents for her arc welding-related enhancements.

Title: A Hybrid Systems Approach: Constraints as Manifolds, Optimization as Computing Connections

The general Kohn-Nerode technique allows suclogical constraints to be converted into continuous constraints. All constraints are then in the continuous domain of a manifold. If the object is to find an optimal control or an optimal solution subject to constraints, this is converted first into the problem of finding geodesics  leading to a goal set relative to a Finsler metric induced by the optimizatiotion criterion, then (above is goal) to the problem of computing by symbolic calculus a connection which gives a control strategy. The connection is measure valued, so actually only a close to optimal control strategy (solution to the optimization problem) is obtained, which is a piecewise geodesic.

Brief Bio:

Professor Anil Nerode is the director of Center for Foundations of Intelligent Systems and the former director of Mathematical Science Institute. Over the last fourty years, Prof. Nerode has done fundamental work in applied mathematics and theoretical computer science: from automata, logic, symbolic computation to constraints, control, modeling and simulation. Prof. Nerode is the pioneer of the research in Hybrid System which becomes one of the most active areas in computation and control. Prof. Nerode has consultancies for both military and industry, serves various national academic commitees, and edits dozens of top journals.

Title:  Planning and Optimization of Dynamic Systems: A Systems Approach

Abstract:

The talk describes an approach for feasible or optimal planning of dynamic systems to attain end conditions in the presence of constraints. The solution of this problem has applications in planning and control of a variety of autonomous systems. The approach presented in the talk is different from the classical solution of this problem using Lagrange multipliers. Here, the structure of a dynamic system is used to explicitly embed the state equations into the cost functional and often eliminate the need for Lagrange multipliers. This approach can be applied to classes of linear  and nonlinear systems such as ‘feedback linearizable’ or that have a ‘differentially flat’ form. The talk will present the theory along with experiments. Also, a number of open questions in the development of higher-order approach will be discussed.

Brief Bio:

Dr. Sunil K. Agrawal is an associate professor in the Department of Mechanical Engineering at the University of Delaware and the director of Mechanical Systems Laboratory. Prof. Agrawal has published a book on Optimization of Dynamic Systems and has written over fourty journal papers on robotics and control. He serves on the Editorial Board and Program technical committes of a number of important conferences in the area. He is a Presidential Faculty Fellow and Principle Investigator for NSF funded research.