http:^^www.cs.utexas.edu^users^qr^abstracts.html

This data set contains WWW-pages collected from computer science departments of various universities

SQUID, which uses SQSim and MSQUID as components, is a system
identification method that refines an imprecise model using a stream
of observations from a physical process.  SQUID uses refutation to
rule out portions of the model space that are inconsistent with the
observations.  We show that this approach to refinement is
significantly more efficient than parameter estimation for models with
functional uncertainty and that it provides greater robustness in the
face of uninformative observations.  <p>

<hr>

<H2><a name="WYLee">Spatial Semantic Hierarchy for a Physical Mobile Robot</a></H2>


Wan Yik Lee.  1996.  
<a href="file://ftp.cs.utexas.edu/pub/qsim/papers/Lee-PhD-96.ps.Z">
<B>Spatial Semantic Hierarchy for a Physical Mobile Robot</B></a>.
Doctoral dissertation, Department of Computer Sciences, The University of Texas
at Austin, December 1996.

<H3>Abstract</H3>

This dissertation describes research to extend and improve the
<i>Spatial Semantic Hierarchy</i> (SSH) approach to robot exploration
and mapping, and to demonstrate and evaluate its effectiveness in
controlling <i>physical</i> mobile robots.  <p>

The SSH approach for robot exploration and mapping was first developed
in the context of a simulated robot, NX, and tested in simulated
environments with very simple models of sensorimotor error.  Physical
implementations of aspects of the SSH approach have been built by
other researchers but they do not provide adequate demonstration of
its strengths or adequate analysis of its conditions of applicability.
<p>

The dissertation work extended and improved the SSH mapping theory
from its original prototype to a version capable of handling
<i>real</i> sensorimotor interaction with a <i>real</i> (office)
environment. The underlying goal of this research is to demonstrate
how symbolic representations and symbol-based behaviors of an
autonomous robot can be grounded in non-symbolic, continuous
sensorimotor interaction with a real environment through the SSH
approach.  The extended theory is implemented on a physical robot to
explore a previously unknown environment, and to create an SSH spatial
description of the environment. This dissertation describes the
improved SSH mapping theory, the details of its implementation on a
physical robot, and a demonstration and evaluation of several features
of the implemention. <p>


<hr>

<H2><a name="WoodWLee">A Qualitative Simulation Based Method To Construct Phase Portraits</a></H2>


Wood Wai Lee.  1993.
<b>A Qualitative Simulation Based Method To Construct Phase Portraits</b>
Doctoral dissertation, Department of Computer Sciences,
The University of Texas at Austin.  <p>

<H3>Abstract</H3>

 We have designed a qualitative simulation based method to construct
phase portraits for a significant class of systems of two first order
autonomous differential equations.  It is intended as a step toward
automated understanding of continuous physical systems.  Differential
equation models are powerful tools for reasoning about physical
systems, but they typically require precise information about systems.
Recently developed methods for qualitative simulation make it possible
to predict all possible behaviors consistent with a state of
incomplete, qualitative knowledge of the world, expressed as a
qualitative differential equation (QDE).  However, qualitative
simulation can fail due to intractable branching, and spurious
predictions.  The field of nonlinear dynamics has introduced the phase
portrait representation as a powerful tool for the global analysis of
nonlinear differential equations.  A state of the system is
represented by a point in phase space; its behavior over time is
represented by a trajectory.  When the phase portrait is
two-dimensional, the solutions to a differential equation can be
characterized by the system's fixed points, bundles of adjacent
trajectories (called flows), and certain bounding trajectories.
Numeric methods for constructing phase portraits require numerically
specific information about the system.  We demonstrate a method and an
implemented program, QPORTRAIT, that constructs two-dimensional phase
portraits from QDE's.  Starting with the total envisionment (a finite
transition-graph representation of the possible behaviors of a
system), QPORTRAIT progressively identifies, classifies, and combines
features of the phase portrait, abstracting away uninteresting
distinctions, and filtering out inconsistent combinations of features.
Because each step in the analysis is validity-preserving, the
prediction is guaranteed to cover all real phase portraits consistent
with QDE.  In its current form QPORTRAIT phase applies to a restricted
but nontrivial set of QDE models.  It requires that all fixed-points
be non degenerate, and be at landmark values for the phase variables.
QPORTRAIT has produced tractable results when applied to qualitative
generalizations of several well-known nonlinear systems.  Guaranteed
coverage of the behavior of a qualitatively described set of QDE's
complements the precision of numeric methods based approaches.  <p>


<hr>

<H2><a name="Pierce">Map Learning with Uninterpreted Sensors and Effectors</a></H2>



 David M. Pierce.  1995.  
<a href="file://ftp.cs.utexas.edu/pub/qsim/papers/Pierce-PhD-95.ps.Z">
<B>Map Learning with Uninterpreted Sensors and Effectors</B>.</a>
Doctoral dissertation, Department of Computer Sciences, The University of Texas
at Austin.

<H3>Abstract</H3>


This dissertation presents a set of methods by which a learning agent,
called a ``critter,'' can learn a sequence of increasingly abstract and
powerful interfaces to control a robot whose sensorimotor apparatus and
environment are initially unknown.  The result of the learning is a rich,
hierarchical model of the robot's world (its sensorimotor apparatus and
environment).  The learning methods rely on generic properties of the
robot's world such as almost-everywhere smooth effects of actions on
sensory features. <P>

At the lowest level of the hierarchy, the critter analyzes the effects of
its actions in order to define control signals, one for each of the robot's
degrees of freedom.  It uses a generate-and-test approach to define sensory
features that capture important aspects of the environment.  It uses linear
regression to learn action models that characterize context-dependent
effects of the control signals on the learned features.  It uses these
models to define high-level control laws for finding and following paths
defined using constraints on the learned features.  The critter abstracts
these control laws, which interact with the continuous environment, to a
finite set of actions that implement discrete state transitions.  At this
point, the critter has abstracted the robot's world to a finite-state
machine and can use existing methods to learn its structure. <P>


<hr>

<H2><a name="Raman">Qualitative Reasoning about Dynamic Change in the
Spatial Properties of a Physical System</a></H2>



Raman Rajagopalan.  1995.  
<a href="file://ftp.cs.utexas.edu/pub/qsim/papers/Rajagopalan-PhD-95.ps.Z">
<B>Qualitative reasoning about
dynamic change in the spatial properties of a physical system.</B></a>
Doctoral dissertation, Department of Computer Sciences, The University
of Texas at Austin.  <P>

<H3>Abstract</H3>

Spatial reasoning is an essential part of human interaction with the physical 
world.  Of the many models that have been developed to support automated
spatial reasoning, most rely on numerical descriptions of a spatial scene.
This dissertation addresses problems where only qualitative descriptions 
of a spatial scene are available, such as natural language understanding,
qualitative design, and physics problem-solving. <P>

We provide the first set of solutions, given only a qualitative description 
of a spatial scene, for reasoning about dynamic change in both the spatial 
and non-spatial properties of a physical system.  We use diagrams to 
compactly input the spatial scene for a problem, and text to describe 
any non-spatial properties.  To match diagram and text objects so their
descriptions can be integrated, we have developed a method for describing 
the conceptual class of objects directly in diagrams.  Then, diagram and 
text objects can be matched based on their conceptual class. <P>

The given problem is solved through qualitative simulation, and
all spatial reasoning is done with respect to an extrinsic Cartesian 
coordinate system.  We model the relative positions of objects through 
inequality constraints on the coordinates of the points of interest.  Changes 
due to translational motion are detected by noting changes in the truth
values of inequality constraints.  We model the orientation of an object 
through knowledge of its extremal points and its qualitative angle of 
rotation with respect to each coordinate axis.  This model has been used to 
reason qualitatively about the effects of rotational motion, 
such as changes in the area projected by one object onto another. <P>

We have implemented our spatial representation as production rules
and as model fragments in the QPC qualitative modeling system.  The former
has been used for solving static-world problems such as understanding 
descriptions of an urban scene.  The latter has been used to reason 
about situations where changes in spatial properties play a critical role, 
such as the operation of transformers, oscillators, generators, and motors.
To support dynamic spatial reasoning, we have expanded the modeling 
capabilities of QPC to include methods for modeling piecewise-continuous 
variables, non-permanent objects, and variables with circular quantity spaces. <P>


<hr>

<H2><a name="Throop">Model-Based Diagnosis of Complex, Continuous Mechanisms</a></H2>


David Rutherford Throop.  1991.
<a href="file://ftp.cs.utexas.edu/pub/qsim/papers/Throop-PhD-91.ps.Z">
<b>Model-Based Diagnosis of Complex, Continuous Mechanisms</b></a>
Doctoral dissertation, Department of Computer Sciences,
The University of Texas at Austin.  <p>

<H3>Abstract</H3>

 In diagnosis, when a hypothesis proposes a variable's value, several
different lines of evidence may be considered; the different evidence
must be arbitrated.  The result of this arbitration consists of a
single best estimate of the variable value and of a measure of that
estimate's plausibility.  The plausibility measure reflects the degree
of agreement among the lines of evidence.  <p>

This report describes HEATX, a program for model-based diagnosis of
non-linear mechanisms with continuous variables.  Previous work in
model-based diagnosis has avoided arbitrating numeric evidence, often
by representing continuous variables as discrete symbols (e.g., high,
cold).  Such restricted representation have had difficulty in
diagnosing mechanisms with feedback or reconvergent fanout.  HEATX
represents numerical data explicitly in the hypotheses and in the
inferencing procedures; it is thereby able to arbitrate evidence
numerically.  <p>

HEATX uses both nonlinear numerical simulations and approximate linear
models to perform diagnosis in the domain of heat-exchanger networks.
The response of these networks to changes in their inputs is
nonlinear; the networks also have feedback and reconvergent fanout.
This dissertation introduces several novel techniques for diagnosing
such networks.  It interleaves the generation of complete fault
hypotheses with several tests on partially formed hypotheses.  Two of
these tests are the qualitative filter and the clustering filter.  <p>

The qualitative filter analyzes the signs of gains between fault and
symptom variables.  The clustering filter constructs linear
approximations to individual components and assembles these into a
linear model of the network.  It then uses the linear model to assess
the consistency of a hypothesis.  It does so by determining whether
there is a value for the candidate fault variable which is consistent
with the quantitative values of the symptom variables; the degree of
agreement between the symptoms and best value for the fault variable
is used to score the hypothesis.  This filter is extended to
multi-fault diagnosis, in which values for several fault variable may
be estimated and judged simultaneously.  <p>


<hr>

<hr>