Newsgroups: comp.robotics
Path: brunix!news.Brown.EDU!noc.near.net!howland.reston.ans.net!vixen.cso.uiuc.edu!uwm.edu!cs.utexas.edu!uunet!psinntp!gdstech!gdstech!wlim
From: wlim@gdstech.GRUMMAN.COM (Willie Lim)
Subject: SPIE Mobile Robots VIII
Message-ID: <WLIM.93Sep14093125@gdstech.GRUMMAN.COM>
Sender: wlim@gdstech.grumman.com (Willie Lim)
Organization: Grumman Corporation, Bethpage, New York, USA.
Date: Tue, 14 Sep 1993 14:31:25 GMT
Lines: 107

This is a report of SPIE Mobile Robots VIII held last week in Boston.
The conference is usually held in late October or early November in
Boston.  Next year it will be from October 31, 1994 to November 4,
1994.

First some opinions: I started attending the Mobile Robots conference
about 5 years ago after hearing good reviews from a friend who used to
be very active (and is known) in the subsumption architecture area.
He thought that for robot hackers the interaction at the conference is
better than those at IEEE RA or AAAI.  In fact many of the early
subsumption work was published at this conference.  Those who live in
the "develop products or perish" world rather than the "publish or
perish" world would find the conference more useful.  I also find that
there is usually at least one paper from MIT presented there.  You
will also find pre-IEEERA or pre-AAAI papers presented there.  That
is, some of these SPIE papers end up in one form or another in the
IEEERA or AAAI conferences.  I also notice, sadly, that those who have
made it to "prime time," that is "graduated" from SPIE Mobile Robots
to IEEERA or AAAI, seem to forget their SPIE roots and don't come
back.  Anyway I feel that the SPIE Mobile Robots conference is a good
way to keep track of on-going work in industry (big companies like GM,
Martin-Marietta, Grumman as well as smaller ones and new startups such
as RedZone Inc., MacLeod Technologies, Inc.), universities (MIT,
Oxford, U. of Colorado, Tsinghua U.), and government labs (e.g.  Naval
Command Control and Ocean Surveillance Ctr. or NCCOSC).  The reason for
my mentioning the above is that many robot hacker type tend not to
know about or ignore SPIE Mobile Robots.  It is a good place to talk
with the people who actually do the work e.g.  researchers/engineers
from industry, innovators from startups, and grad students or visiting
scientists from universities.  If you want to find out how they get
their robots/sensors/actuators/software/etc working, what "tricks"
they used, or what they are currently working on, this is the
conference to hang out. :-) This is also the place to sell your
products or yourself since many of the industry product developer
types are here too.


WORKING GROUP ON ROBOTICS AND MACHINE PERCEPTION: This meeting was
chaired by Paul Schenker (JPL).  There were about 20-30 people in the
meeting including David Casasent (CMU) the 1993 SPIE President and the
annual Conference Chair of SPIE Intelligent Robotics under which are
several conferences including Telemanipulator Technology and Space,
Sensor Fusion, and Mobile Robots.  There were some discussions on what
format the conference should take.  For example whether conference
proceedings should be available at the start of the conference or
whether to shift the focus.  Unlike other conferences, SPIE papers are
accepted based on the submission of an abstract.  The cons for this
include people not delivering papers or showing up once the abstracts
are accepted.  The big pros is of course people get to include in the
latest result right up the day of conference.  Also some people use
this conference as a dry run for IEEERA and AAAI, so it the place
where the results are first reported.  Someone also mentioned that
SPIE tend to attract a bigger foreign participation.  An SPIE
connected person said that this because SPIE had worked very hard to
get foreign participation.  Other discussions include making the SPIE
abstracts and other publications available on-line.  They are looking
for people to help out by joining the working group.  Email
spie@mom.spie.org for more information.

SPIE EXHIBITION: Most the exhibits were on optics and image
processing.  However TRC had a booth there.  They displayed their
LightRanger (LIDAR: Light Detection and Ranging) system with an Intel
386/33 host.  It has a range of 9 m.  There is also an info sheet on
their LIDAR based beacon navigation system.  The thing that caught
people's attention is their VME or PC bus BiSight Microminiature
system (robotic binocular camera platform).  Their demo showed the two
pencil size cameras (cameras not included) mounted on that system
tracking a light.  It was fast and responsive.  They have sold several
of the system to universities for active vision research.  They also
have a TRC labmate and the sonar and IR based proximity subsystem on
display.  The proximity subsystem is actually a 68HC11 based card with
an interface card attach.  The interface card is capable of
controlling 8 sonar or 8 IR sensors.  You can have up to 24 sonar and
24 IR sensors (viz. proximity subsystem plus 2 additional interface
cards) in th whole configuration.

SPIE MOBILE ROBOTS VIII:

The most interesting paper is "Sozzy: A Hormone-Driven Autonomous
Vacuum Cleaner," by M. Yamamoto, visiting scientist of Rob Brooks'
group at the MIT AI Lab.  He talked about giving the robot emotions:
sad, happy, and aggressive.  It is sad when it loses sight of its home
beacon.  When that happens it stays still for a while and start
looking for the nearest human to get help.  It is happy when it is
eating dirt (vacuuming).  If it gets blocked or trapped and can't
move, it will start bumping into "obstacles" in the hope that some
these will give and let it escape from the trap.

Other papers include a laser based navigation and control system (E.W.
MacLeod of MacLeod Technologies, Inc.), a robot for automatic
harvesting of asparagus (Istituto Elettrotecnico Nazionale, Italy),
coordinated control of multiple security robots (G. Gilbreath,
NCCOSC), "How to communicate with zillions of robots" (Doug Gage,
NCCOSC), "Lessons learned on a low-cost GNS for the SSV Program"
(Benny Gothard, Martin Marietta), and "Autonomous Long Life Vehicle"
(Alex Peck, Grumman Corp.).

(-: Footnote: Ten of us went to a restaurant in Boston's Chinatown and
had an authentic Chinese feast.  I did the ordering and everybody had
to obey the "just eat, don't ask" rule.  This may become an annual
thing.  The MIT hacker "chopsticks only or starve" rule was not
enforced.  :-)




Willie
