|
|
Elmer the Tortoise
In 1949 Grey Wlter built his first ‘model animal’ – Elmer the tortoise - using only two electronic brain cells
By Owen Holland
For me, it all began several years ago at a conference. As I queued for lunch, the man next to me, a Hungarian, remarked, “Ah! You’re from England. I worked there some years ago, at the Burden Institute in Bristol. Do you know it? That’s where the famous neurologist Dr. Grey Walter built his robot tortoises.”
I can’t remember what else we talked about, but his words somehow stuck in my mind. I knew that in 1949 Grey Walter had built a robot to demonstrate his ideas about how the brain worked. He did not think humans were intelligent just because they had ten billion brain cells, but rather because their brain cells were connected up in many different ways. So he built his first ‘model animal’ – Elmer the tortoise - using only two electronic brain cells, connected together in several different ways. Not much of a brain, but Grey Walter had designed it very cleverly. Elmer would explore a room, looking for lights, moving towards them, circling them, and then wandering off in search of more. If he found a mirror, he would do a dance in front of it; if he came across his sister, Elsie, he would dance with her. If he came across an obstacle he would try and push it out of the way; if this didn’t work, he would go round it. And when his battery began to run down, he would return to his hutch, and plug himself in to his power socket, setting off again in search of lights when his battery was fully charged. Grey Walter had proved his point – two richly connected brain cells were enough.
As computers and artificial intelligence developed, scientists began to develop complex robots very different from Grey Walter’s, and soon the tortoise was just history. However, in the 1980s, the complex artificial intelligence approach stalled, and researchers began to make simple robots, imitating insects rather than humans. I was part of this movement, and in 1993 I helped to found a robotics laboratory at the University of the West of England, Bristol – two hundred metres from the Burden Institute! Of course I called in to see whether any tortoises still survived, but I was told the answer was no – Grey Walter had died in 1977, and all the tortoises had been lost or destroyed. Seeing my interest, they kindly gave me access to their records, and one day I found a document proving that when Grey Walter designed the tortoise, he had used exactly the same techniques that we were using forty years later. This gave the tortoise a new importance – it was the ancestor of our modern line of robots – and I decided to gather enough information to have an exact working copy made. I began to track down and interview his former colleagues, looking for photographs and other records. Months later, one of these asked me, “Have you talked to his son? I have his London phone number from about ten years ago.” Without holding out much hope, I rang the number. It was answered promptly: “Hello, Nicolas Walter speaking”. I explained that I was trying to gather details about the tortoises, and asked him if he had any photographs or other details. “I can do better than that,” he said; “I’ve got my father’s tortoise. It’s been in the cellar for twenty years, in its hutch.” Next day it was in our laboratory, the day after that it was working again, and later this year it will go on permanent public display in the Science Museum in London – a fitting end for one of the most important robots ever built.
Owen Holland is a professor of computer science at the University of Essex in the UK. He is most famous for his work in biologically-inspired robotics.
Teilen
Für 94 Franken pro Jahr gibt es NZZ Folio auch im Abonnement. Näheres hier.
Urheberrecht gilt auch im Internet: Verlinken erlaubt, Kopieren verboten.
|
|
|