|
|
Computer Science Assistant Professor
rpastel at mtu dot edu
204 Rekhi CS Hall |
| |
|
|
| |
|
| |
| Primarily, my
research is in the broad area of Human-Computer Interaction, HCI. I
choose to study HCI deliberately. Foremost HCI combines artistic sensibility
with mathematical precision. An interface designer must use programming
languages to create an interface to the computer which is both efficient
and pleasant for the user. But HCI is more than static art work; a user
interface is intended to have utilitarian value and interact with both
the user and the computer. Consequently, the interface designer should
understand the user’s goals, and their cognitive processes. Finally,
HCI is an under appreciated field in academic computer science but not
in industry. I believe there are many opportunities in HCI research
and industry. |
| |
 |
|
HCI research can be divided into three broad areas: interface
design and implementation, interface software and hardware tool development, and interface
theory and evaluation. These three areas (theory, design and tools)
form the three supports that all user interface must be based on to be
successful. My research spans all three areas. |
| |
| HCI is a particularly approachable
research area because using the computer we all are exposed to many user
interfaces, and we naturally criticize and evaluate them. The interface
that most of us use most of the time is the desktop environment.
The desktop environment is an essential interface for personal computers;
it should act as a natural interface to the operating system and file
system, besides providing basic utilities. I have been frustrated with
all desktop environments, so naturally, a redesign of the desktop was
my first endeavor. The paper “Desktop Environment
as a Problem Solving Tool,” considers the design of a zoom desktop
that uses a graph data structure to associate windows. |
| |
 |
|
Because of their portability, personal digital assistants
are attractive platforms for users that benefit from mobility, for example
conducting laboratory or field experiments. Nathan Skalsky and I proposed
a PDA-based science
laboratory, PBL, and designed Experimental Assistant an application
for the TRIcorder. |
| |
| Although I do not have much time
to play computer games, I would enjoy moding or developing computer
games. I believe that computer games push the frontier of computer technology
and demonstrate new possibilities for more utilitarian HCI. Interface
designers must consider that many future computer users are introduced
to the computer via games and therefore are acquainted with alternative
interfaces. I believe that a fun research area is interfacing open source
game engines to common applications. |
| |
| Computer mice and touch pads are
unique peripheral devices and perform an essential function in HCI. They
provide rapid and intuitive user input to the computer. The mouse was
invented in 1964 by Engelbart, and by 1981 became popular in WIMP (windows,
icons, menus and pointer) interfaces that have become ubiquitous in user
interfaces. Pointing devices have developed little from point and click
inputting, selection utility. Seldom do interfaces use direct manipulation
techniques, for example drag and drop manipulation of icons. Direct manipulation
techniques begin to make use of the pointer’s full potential by
incorporating the pointer’s motion to interpret the user’s
intent. Gesturers are natural human acts and interfaces using gestural
input are potentially more humane and efficient. I believe that developing
gestural interfaces is a fruitful HCI research area. |
| |
 |
|
The personal digital assistant, PDA, is a challenging platform
for interface design because of the restrictions imposed by the limited
screen size and lack of keyboard. Gesturing interface can overcome many
of the PDA challenges. In the paper “Demonstrating
Information in Simple Gestures” I introduced the Simple Gesturing
User Interface, SGUI, an application programming
interface, API, for writing gestural interfaces on the palm pilot. SGUI
is fast and light weight gesture interpreter by tracking only the gesture
bound box. I would like to port SGUI to Java. |
| |
| Kyle Erickson, a student,
compared alternative gesture interpreter, and programmed one of the interpreter that classifies the gesture
by tracking the gesture through 9 regions of the gesture’s bounding
box.
Although the mouse appears to be under direct control of the user,
mouse motions are interpreted by the drivers and the operating system
before the cursor is moved on the screen. Seldom is the cursor motion
linear with the mouse motion. Frequently the cursor is accelerated,
for example moved trice as far, for large mouse motions and not accelerated
for small mouse motions. This mouse-cursor transformation allows the
user to move the cursor quickly for gross motions while still retaining
accurate cursor control for fine motion. This separation of cursor from
the mouse allows for the intelligent interfaces to assist the user.
The interface knows the location of buttons; it can accelerate the cursor
towards the button. A student, Chris Fuller, wrote the Java GravityMouse class and
a test platform for GravityMouse.
I believe that the basic gravity mouse is useful for users with disability
using interfaces with a small number of buttons.
Desktop environment development in linux requires modify XFree86 and
installing the new version. Chris Balzek developed procedures for running multiple
versions of XFree86 on the same computer. Student installation of
XFree86 required that Chris set up and maintain the first HCI lab at
MTU. Chris has also prepared linked
documentation web pages for the XFree86 source code.
Typical touch pads can locate only a single area on the pad because
the electrodes are arranged in columns and rows that extend across the
pad. Dr. Westerman developed the software for interpreting gestures
on pads that use inputs from a grid of electrodes. More than a single
area can be determined; in fact a proximity image of the both hands
is created. FingerWorks has developed
a touch pad that can be simultaneously used as a keyboard a mouse. I
believe that this device can be used for many applications. |
| |
 |
|
Designing Experimental Assistant, a graphing application
using SGUI, to use gestures required new design approach. Particular attention
needed to be focused on the action of the interface objects. I called
this design model object-action
association. |
| |
| There are few quantitative
measures in HCI to determine interface effectiveness. One of the few
quantitative measures is the time to perform tasks on the computer.
A simple task is moving the cursor to a button. Many computer scientists
have measured the time to move the cursor to a button, and confirmed
that the time is governed by Fitts's law. The Fitts's
law states that the time to move to a button is logarithmically
proportional to the initial cursor target separation divided by the
target width. Joe Vaillancourt developed a test platform
for verifying Fitts's law.
Johnny Accot and Shumin Zhai at IBM developed the steering law,
which extends Fitts’s law to guiding the cursor through a prescribed
path on the screen. The steering law and the accompany theory provides
a means for measure the efficiency of gesturing interfaces. Robert Moore,
Jake Camplin, and Scott Gross have developed a test platform for verifying the steering
law. Dmitry Podkuiko is currently administrating steering law tests
and Dr. Struther is analyzing the data. |
| |
| Because a large part of my job
is teaching, data structures and HCI, I have the opportunity to conduct
research in computer science education, and the computer science education
research has generated new ideas for HCI research. |
| |
| HCI is a neglected area in
computer science pedagogy. The academics are debating what and how HCI
should be taught. I have proposed integrating science and research in an HCI design course.
I am now considering developing an HCI course, which focuses on the
web and applications on the web. This focus will allow HCI students
to learn in detail tools and techniques in popular HCI venue, the web.
|
| |
| Attendees at a publisher’s
symposium were interested in the development of group labs for the introductory
data structures course. Because I have developed physics group labs, I
thought I am qualified to develop group labs for the data structures course.
I have written 4 data structures group labs and this year I will evaluate
their pedagogical effectiveness. |
| |
| |
- Port SGUI to Java
- Analyze Gravity Mouse user testing results and develop Gravity
Mouse algorithm
- Administer group labs and analyze results
- Develop web profiling techiques and tools
- Converting test platforms to applets
- Administer steering law tests and analyze results
|
| |
| If you are interested in any
these research areas, or have research ideas of your own in HCI or computer
science education, please see me. Undergraduate students can earn independent
study credits and and graduate students can earn research credits. |
|
|
|
|
|
|
|