Motion Planning via Relative Equilibria for Autonomous Vehicles
F. Bullo*
University of Illinois; National Science Foundation, CMS-9502224

The focus of this research is on trajectory planning methods to be used in autonomous underactuated vehicles. Mode-switching in flight control and orbit transfer in aerospace navigation are instances of this problem. Recently, great progress has been made on designing a control Lyapunov function that can stabilize a class of steady motions called relative equilibria. These motions arise in the context of mechanical systems with symmetries. Relying on these advances, this project focuses on how to design a provably stable method of switching between different relative equilibria. The aim is to develop techniques amenable to on-line implementation. A second direction of research is to understanding relative equilibria and their stabilization in the context of flight control problems with general aerodynamic forces.

Motion Primitives for Stabilization and Control of Underactuated Mechanical Systems
F. Bullo,* N. E. Leonard*
University of Illinois; National Science Foundation, CMS-9502224

This project focuses on the design of motion primitives as algorithm building blocks for stabilization and control of a class of mechanical systems that includes spacecraft, submersibles and hovercraft. The underactuated systems have a invariant-Lagrangian dynamics, and this structure is exploited in control design. Various systems are examined in terms of their controllability properties. This analysis leads to the design of motion primitives that generate motion in arbitrary directions. These primitives can then be used for a variety of low velocity maneuvers, such as point stabilization and static interpolation. These results show how these vehicles can perform tasks such as short range motions and hovering despite being underactuated.

Developing a Distributed Software Emulation for the RAMP Flexible Manufacturing Systems
W. J. Davis,* A. L. Brook
U.S. Army Rock Island Arsenal, Batelle TCN95300

Using a new object-oriented simulation framework, a distributed software emulator is being developed for the rapid access to manufactured parts (RAMP) flexible manufacturing systems (FMS) operated by the Department of Defense. The emulator models by defining distributed central objects for each of the 70 controllers. The controllers interpret with a client server. The same server also manages the Java-based controller allowing the emulation to be viewed anywhere on the World Wide Web. The project will eventually seek to implement an on-line intelligent control architecture RAMP FMSs.

Modeling and Analysis of Large-Scale, Discrete-Event Systems
W. J. Davis*
University of Illinois

A new hierarchical framework for the intelligent control of large-scale, discrete-event systems has been formulated. The conceptualized hierarchy defines a coordinated object that can be employed recursively to describe the desired hierarchy. In order to distribute planning and control, an intelligent controller is included within each coordinated object. The essential mechanisms to coordinate the distributed planning and control are now being explored. New object-oriented simulation tools are also being developed to model the interaction among the coordinated objects.

Orthogonal Load Resolution in the Control of Multilink Systems
E. N. Kuznetsov*
University of Illinois

A general load on a multilink system resolves uniquely into two orthogonal equilibrium and perturbation load components, the former equilibrated by the system, the latter causing its motion. In this motion, the perturbation load is counteracted not only by the system inertia but also by the stabilizing effect of the equilibrium load. Thus, the equilibrium load is not just wasteful (requiring energy to generate it) but outright counterproductive. Only a control action producing pure perturbation loads can achieve full power efficiency; for any multilink system this action is unique, with a certain number and arrangement of actuators.

Advanced Variable Structure Controller for Multivariable Nonlinear Systems
J. V. Medanic*
University of Illinois

This research considers design of a nonlinear, variable structure controller for multivariable nonlinear systems. The controller is based on the concept of polar partition of the polar control and is intended to insure semiglobal stability, to be robust to plant perturbations, and to possess qualities of interest in such applications as fast prototyping and off-the-shelf availability.

Fault Tolerant and Reliable Control Systems
J. V. Medanic*
University of Illinois

The need to continuously operate technical systems in the presence of structural changes in the system, damage to parts of the system, or failures in the control system has long been recognized as an important issue in system design and operational control. The project emphasis is on the development of systematic procedures for the design of fault-tolerant control systems. The focus is on design of reliable control systems using redundant sensors and actuators, on fault assessment methods, reconfiguration procedures and development of controllers based on coarse partitioning of the state space, using methods based on variable structure control, cell-to-cell mapping, and state space partitioning.

Nonlinear Integral Transformations for Control of Nonlinear Systems
J. V. Medanic*
University of Illinois

The purpose is to develop the concept of nonlinear integral controls for nonlinear dynamic systems based on integral transformations. The approach applies an integral transformation to reduce the system to a linear system or to a nonlinear system with a simpler structure. The method is used to resolve the stability problem and the tracking problems for classes of nonlinear systems, as well as to provide other aspects of desired performance. The method relies on the integration of nonlinearities (a consequence of the integral transformation), as opposed to differentiation of the nonlinearities that accompany most methods proposed and developed in the past.

Reconfigurable Systems for Tailless Fighter Aircraft
P. G. Voulgaris,* J. Medanic, W. R. Perkins, P. R. Kumar, T. Basar
McDonnell Douglas Aerospace (Conducted in the Coordinated Science Laboratory)

The objective in this project is the development and evaluation of reconfigurable control algorithms for the new generation of tailless fighter aircraft. The main effort is placed on the provision of system identification algorithms that are reliable, fast for on-line implementation, and efficiently support reconfigurable dynamic inversion control laws. The overall reconfigurable system is tested and evaluated on nonreal-time simulation as well as real-time piloted simulation using the McDonnell Douglas facilities.

Development of Virtual Prototyping Systems
R. S. Sreenivas,* W. R. Norris
Caterpillar, Inc.

The notion of "steering quality" is difficult to quantify in the design of steering systems for earth-moving vehicles. We envision an expert driver operating a simulated vehicle within a virtual environment where the parameters in the design of a steering system can be altered instantaneously. In this paradigm, the process of trial-and-error design becomes a viable option. This project involves the derivation of vehicle models of appropriate complexity and detail that can be simulated in real-time within a virtual environment. To improve the real-time performance of these models, particular attention is focused on artificial neural networks.

Modeling, Analysis, Control, and Performance Evaluation of Discrete Event Dynamic Systems
R. S. Sreenivas*
National Science Foundation, ECS-9409691

We concern ourselves with 11 broad research issues in the modeling, analysis, control, and performance evaluation of Discrete Event Dynamic Systems (DEDS). Research in modeling and analysis concerns the identification of modeling paradigms suitable for supervisory control of DEDS. The control of DEDS concerns the archetypal problem of elimination of deadlocks. The performance analysis of DEDS concerns an approach that will permit a system designer to modify automatically discrete event simulation programs that only estimate performance so as to efficiently obtain the sensitivity of the performance. This sensitivity information can be used toward system optimization using stochastic approximation techniques.

Determining Hydrogenerating System Stability and Performance
L. Wozniak*
University of Illinois

Hydrogenerator governors are designed with predetermined rotational inertias and conduit dimensions for maximum acceptable off-speeds (speed deviations from reference). However, this parameter selection may not be favourable for stable operation and satisfactory small signal level performance when governing isolated loads. Poorly governed plants operating in interconnected systems degrade the overall stability. This work develops a graphical method to determine expected performance and stability characteristics based on inertia and conduit sizing decisions. It is directed toward mechanical and civil engineers involved in the design and associated economics of plant layouts.

Efficiency-based Optimal Control of Kaplan Hydrogenerators
L. Wozniak,* P. Schniter
University of Illinois

This research investigates an optimal strategy for controlling the speed response of Kaplan hydrogenerating systems to decreases in load. Typically, primary control gates restrict and redirect water through the turbine to stabilize and transfer the system to operating point demand. The adjustable turbine blade angle is used to return to maximum operating efficiency at the new load level. The overspeed reduction is limited by the conduit's ability to withstand the overpressure caused by the flow restriction at the turbine. A control scheme using gates and blades simultaneously and independently is developed.

A Graphic Approach to Hydrogenerator Governor Tuning
L. Wozniak*
University of Illinois

A number of published works deal with governor tuning for speed control of hydrogenerators. This work is based on the hypothesis that some system parameters are not known at the design stage. It develops a graph that can be used to predict optimum proportional and integral gains based on four parameters: the time constants of the water column and the rotor inertia and the self-regulation constants of the turbine and the loading grid. The pole cancellation method of design is used and the results are posed in an easy-to-use format not requiring the solution of systems of equations.