I'm an associate professor in the computer science department at North
Carolina State University. My Ph.D. is from the University of
Massachusetts, Amherst; my CV and
papers are online. (For the less technically inclined, I've
written several informal essays about interaction
design.)
Research in my lab can be summarized as targeting models of
interaction, drawing on concepts in artificial intelligence,
human-computer interaction, and cognitive science. Over the past
several years I have become interested in the concept of tools,
both physical and abstract.
Random bits
A few news items and other things of note (Spring, 2008):
I've released the code for simple versions of a partial-order planner and a GraphPlan planner (simple-POP and simple-GP),
which I hope will help others in teaching AI and Lisp courses. (The
AI
Planning Resources page may also be useful.)
Current projects
Robot tool use:
Some tools seem to have the property that one can tell how they should
be used just by looking and experimenting---no instruction or
specialized knowledge is needed. Affordances are part of the
explanation why, as is general tool-using ability. What exactly do
these concepts mean? This project examines how we might build a
habile (tool-using) agent, which we call Canis habilis.
Affordances
and tool use in user interfaces:
The Canis project grew out of work on making interactive
objects in user interfaces behave more like objects (tools) in the
real world. We developed desktop and multi-touch versions of a
drawing application, HabilisDraw and HabilisDraw DT.
A script describes what's going on in the
HabilisDraw DT video.
Image
processing in user interfaces:
Most interface agents interact with applications through an API, which
provides a different perspective on a system than that of the user.
What might we learn from interface agents that can interact with
applications through their user interfaces, just as human users do?
This project addresses perception, cognition, and interaction issues
for these agents, which we call interface softbots.
An offshoot of this effort has led to a project that imposes the
interaction logic of video games on actual video, as in the
proof-of-concept game, AugFrog. (Our interest is more in models of
such games than the polish of implementation.)
Cognitive modeling for HCI:
We have ongoing collaborations with Frank Ritter at the Applied
Cognitive Science Lab at Penn State, and with Wayne Gray at the
CogWorks Lab at RPI, to explore the relationship between models of
hands, eyes, and cognitive processes.
I'll be a bit polemical here: There's not enough attention given to
science and engineering in HCI. (I'm guilty of this failing myself in
much of my pure HCI work.) While the field has good standards for
empirical evaluation of interactive systems, what I think is needed is
a better understanding of the principles of interaction. Cognitive
modeling and formal modeling are the best candidates for producing
such understanding. (Some of my favorite researchers in the field are
Wayne Gray, Bonnie John, and Dave Kieras on the cognitive modeling
side, Alan Dix and Harold Thimbleby on the formal modeling side.)
Software
Simple-POP and Simple-GP
are planning implementations simple enough to present in one or
two classes each (including walking through the code) in an AI or Lisp
programming course.
G2A is a
system that translates high-level procedural GOMSL models into
detailed cognitive ACT-R models.
SegMan
is a perceptual substrate that uses computational vision to
"see" the Microsoft Windows graphical direct-manipulation
interface, developed as part of the interface softbots project. The
latest software release is
downloadable; it allows planners, cognitive models (based on ACT-R,
Soar, and EPIC), and programmatic controllers to interact directly
with off-the-shelf Windows applications.
Last updated, Wednesday, March 20, 2002. Mail questions or comments to stamant@csc.ncsu.edu.