CFP for a special issue of the journal Human-Computer Interaction

Topic: "Multimodal Interfaces"

After the CFP was forwarded to us here at ASEL, Rick sent me an email
saying, "We should grab this". 

Since an intent to submit is due Friday, September 1995, I wrote out
a draft abstract of the proposed paper. This will give us tomorrow to
iron out the kinks. I am appending the abstract to this email.

The working directory is:


	call_for_papers			The cal, for papers
	hci.abs.fm				The maker version of the attached abstract
	hci.abs.txt				The text version of the attached abstract
-- begin --

Title: Multimodal Control of an Assistive Robot

Zunaid Kazi, Matthew Beitler, Marcos Salganicoff,
Shoupu Chen, Daniel Chester and Richard Foulds

Applied Science and Engineering Laboratories
Alfred I. duPont Institute/University of Delaware
Wilmington, Delaware USA


The Multimodal User Supervised Interface and Intelligent  Control
(MUSIIC)   system  involves  the  integration  of  human-computer
interaction with reactive planning to operate a telerobot for use
as  an assistive device. The system is intended to meet the needs
of individuals with physical disabilities to control an assistive
robot in an unstructured environment, rather than in a structured
workcell.  We  describe  a  novel  approach  for  an  intelligent
assistive  telerobotic  system  for  such an environment: speech-
deictic  gesture  control  integrated  with  a   knowledge-driven
reactive  planner  and  a  stereo-vision  system  which  builds a
superquadric shape representation of the scene.

Our approach is based on the assumption that the user's world  is
unstructured,  but  that objects within that world are reasonably
predictable. We reflect this arrangement by providing a means  of
determining  the  three-dimensional shape and pose of objects and
surfaces which are in the immediate environment, and  an  object-
oriented  knowledge  base  and planning system which superimposes
information about common objects in the three-dimensional  world.
A third major aspect involves the multimodal user interface which
interprets  the  deictic  gesture  and  speech  inputs  with  the
objective  of  identifying  the  portion  of  contour  that is of
interest to the user, as well as  allowing  an  easier  and  more
flexible  interface  for  robot  control. Users of our system use
deictic gestures (pointing, achieved  by  a  head  mounted  laser
pointer)  to  indicate locations, and spoken commands to identify
objects and specific actions. The combination of spoken  language
along  with  deictic  gestures performs a critical disambiguation
function by binding the  spoken  words  in  terms  of  nouns  and
actions  to  a locus in the physical work space. The spoken input
is used to  supplant  the  need  for  a  general  purpose  object
recognition  module in the system. In addition, we are developing
a simulation environment that will enable us to  investigate  the
modalities  of  the  human-computer  interaction  in  a  low risk

-- end --

Zunaid Kazi                  kazi@asel.udel.edu (or @cis or @strauss)
CompSci & Robotics                     http://www.asel.udel.edu/~kazi