Wow...researching this one proved to be very interesting and
confusing. What we do know is, the IEEE contest you are referring to
was held June 5-7, 1979 in New York, as the highlight of the National
Computer Conference. There were over 6,000 entries and this was
dwindled down to 15 finalists.
This much is certain. From here, it gets a bit hazy.
It seems that of the 15, only 4 solved the maze (which was 8x8) on
their first run and 3 more made it at their third attempt. Most
references state that the winner was "Moonlight Flash" who used the
wall-hugging technique you described in your question. It was
considered a "dumb" mouse as the other entries were basically using
artificial intelligence (which was the key factor for the race) to
learn the maze, while Moonlight Flash, simply ran a "finger" along the
walls to come out the other end.
I have also seen a reference that states a team from Battelle
Northwest Laboratories built the winning mouse named "Moonlight
Express" and that it used LED's and photodetectors to monitor the
walls of the maze and it was declared the winner, though not the
fastest because an entry used wall-hugging and was deemed ineligible.
According to the Battelle website, the Moonlight Flash and the
Moonlight Special (another entry) finished 1st and 3rd in the "wall
hugger" category. This is the only reference I can find regarding
seperate categories.
James Hamblen of Martin Marietta Aerospace designed the second place
mouse. I have uncovered a Dr. James Hamblen at Ga Tech whose bio
states he worked at MM and he is a professor in the computer
engineering. I sent Dr. Hamblen an email to see if I can uncover the
truth.
The only reference I see to Harvey Wallbanger is withing various
definitions of the programing slang term "wall follower". I asked Dr.
Hamblen if this was his mouse or for at least some more information on
his entry. I also asked if there were any photographs of his mouse,
though on the Battelle page, referenced below, there is a picture of
one of the moonlight mice.
As soon as I hear back from Dr. Hamblen, I will post his reply here.
Though, I tend to think the first reference below, which is reprints
of several magazine articles, is most likely the most correct source
until I hear from Dr. Hamblen.
I thouroughly enjoyed researching your question!
Regards,
-THV
Search Strategy:
"harvey wallbanger" robot
micromouse 1979
"Amazing Micro-Mouse"
"Battelle labs"
James Hamblen
moonlight flash micromouse
References:
ON MICROMICE AND THE FIRST EUROPEAN MICROMOUSE COMPETITION
http://davidbuckley.net/RS/mmouse/AISBQ39.htm
MICROMOUSE
http://www.ee.cityu.edu.hk/~rtbrad/micromouse.html
A bit of Micromouse History
http://www.np.edu.sg/alpha/micromouse/mice_history.htm
PNNL History
http://www.pnl.gov/main/welcome/history.html
The New Hackers Dictionary
http://www.outpost9.com/reference/jargon/jargon_38.html |
Clarification of Answer by
tar_heel_v-ga
on
26 Aug 2004 09:50 PDT
I received the following from Dr. Hamblen:
"Wow, that's a while back - I think mine would be a 78 first time trial race
second place winner but perhaps the official race was 79 and the time trials
started in 78!
I was at the first time trial at Disneyland (just a couple made it out
of the maze) that's where the photo is likely from, one later in LA and
then NYC. I had rechargeable batteries and let the Spectrum guys run the
robot as much as they wanted for demos - so it was in a lot of photos. I was
back at grad school before the NYC race and did not have the time to sink into
having everything in great shape for the last race in NYC and did not
do too well there.
There were a couple articles in IEEE Spectrum about this - just a few
too many years back to find it on the computer databases - I would need to
review them to be 100% on everything. There were contests for several
years and I recall the mouse from Batele winning. There were a number of time
trials and they announced winners (anyone that made it out of the maze)
at each time trial - so that might be part of the confusion. The IEEE guys
would be the final word on it. The guys in the IEEE Spectrum editor's
office kind of ran it. There was a Roger Allan that was the main person
running it all and the official judge and rule maker. I seem to also
recall a fast "dumb" wall hugger at one of the last races. I am not
sure about the
names of the mice and I did not call mine Harvey Wallbanger.
I seem to recall a rule that the mouse the had the fastest time after
learning the maze in three runs was one prize (AI) and the fastest
single run was another.
Articles describing the races & rules appeared in
Science, September 1978, pp. 800-801, (early ones talk about
rules & time trials)
IEEE Spectrum, November 1978, pp. 63-65,
IEEE Spectrum, September 1979, pp. 62-65, (this one might have
the final race results)
and Byte, September 1978, pp. 10-12.
I have copies somewhere at home but I think but it might be pretty hard
for me to find - I will look around tonight. If you could track down Roger
Allan or Donald Christiansen they would have a lot more details and be the
final word. Perhaps the current editors at IEEE Spectrum might know whey they
are now www.spectrum.ieee.org"
I am in the process of trying to contact the folks at IEEE Spectrum
and see what I can learn. This has been a fascinating question!
-THV
|
Clarification of Answer by
tar_heel_v-ga
on
27 Aug 2004 05:54 PDT
Wolfgang..
I received several emails overnight. One was from Roger Allan, the
editor at the time of the race, of the sponsor, IEEE Magazine. He is
reviewing he "volunmious" amount of information on the race.
The other was from Dr. James Hamblen, who I mentioned in my original
answer. He was kind enough to provide me with a scanned copy of the
1979 article covering the race in IEEE Spectrum Magazine. This can be
found at http://users.ece.gatech.edu/~hamblen/papers/micromouse/mmrace.pdf.
WARNING: This file is 24MB. I would recommend downloading the file
and viewing it on your machine instead of online.
In reviewing the article, there WAS an entry named Harvey Wallbanger
and it finished 3rd fastest and was one of only 4 finalists to
complete all 3 runs of the maze. Turns out that Wallbanger was first
wall hugging mouse to finish the maze in less than one minute. The
article has a picture of the Wallbanger on page 4 of the document. It
was designed by a team from Hewlett-Packard.
Again, thank you very, very much for the opportunity to answer this
question. I have answered over 400 questions since becoming a
researcher, and this has been the most enjoyable project I have done.
I am actually considering creating a website based on the information
gathered during my research to provide the correct information on what
I think, based on the time it took place, is a fairly important step
in the growth of computer and AI technology.
Regards,
-THV
|
