Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.milw.wisc.edu!bionet!ig!arizona!gln
From: gln@arizona.edu (Gary L. Newell)
Newsgroups: comp.cog-eng
Subject: Re: "Cognitive queueing" (was: Typing versus Handwriting)
Summary: a couple more thoughts
Message-ID: <12172@megaron.arizona.edu>
Date: 30 Jun 89 08:23:48 GMT
References: <1440001@hp-ptp.HP.COM> <1440002@hp-ptp.HP.COM>
Organization: U of Arizona CS Dept, Tucson
Lines: 30



 I just thought that I'd add a couple more thoughts on this topic.....

1) IBM recently announced a product like the one we talked about (hand held
for use on warehouse floors etc.) - they worked on the area in general with
that goal in mind - I have no details about it and I am not even sure if
they have begun production yet but I saw a little segment on the network
news about 2 months ago. I have yet to see a new research paper from their 
group since that news blip so I assume that something is happening.....

2) Note that recognition of cursive script seems pretty bad at this point
unless the vocabulary is kept quite small. The problem of segmenting the
individual letters appears to be too difficult. Also, many words
are created using a single continuous stroke which complicates things
even more. The fact that many people wait to cross their t's or dot
their i's also makes real-time recognition more difficult.

3) I have a paper around here somewhere called ' A comparitive evaluation
of conventional vs handwriting interfaces' or something like that - I'll try to
dig it out and post the reference - Cathy Wolf comes to mind as the author but
I could be way off.

4) I am currently working in this area and would be very interested in hearing
others results/failures/ideas/suggestions on any area of Gestural interface -
hardware, human factors, recognition or pre/post processing algorithms - 
anything..... 


		Gln