home  >  Olympics 90  >  IAPR Newsletter 21 May 2007

IAPR Newsletter Volume 13 Number 4 February 1991 p2
International Association For Pattern Recognition

The First Robot Olympics

MOST PEOPLE DON'T GET PASSIONATE, obsesive and captivated about building washing machines, telephones or trouser presses. So why is it that as soon as you start building clever robots everybody seems to be fascinnated? To children they are wonderful toys whilst to the non-technical adult they are a combination of something to fear (horror movie or lost job) or some fun gadget for the home. To the business man they represent a technology that he not afford to ignore if his industry is to have a future, whilst to the TV cameraman they simply make good visuals. To the politician they may one day have the vote, to the military mind they might fight wars, whilst hopefully leaving humans out of it. For the handicapped they represent a liberating technology which allows them scope for expression which previous generations missed. For the robot builder they are a goldmine of information showing how others have chosen to build legs, arms, sensors and control systems, and for the academic, something to make experiments with or write about. For the philosopher they challenge and shift our ideas about intelligence (or stupidity) and for the religiously minded, deep concern about the future of the human race.

Richard 1st, a prototype anthropomorphic robot head being developed at the Turing Institute, which acted as commentator during the Robot Olympics.

As was clearly apparent by the capabilities displayed at the First Robot Olympics, all these ideas were well wide of the mark. Over sixty robots from twelve countries gathered in Glasgow at an event organised by the Turing Institute. As with any new and emerging technology, the results are typically unreliable, sometimes unpredictable and fraught with technical problems. If anything was surprising, then it was the fact that virtually all the robots managed to work properly during 27th and 28th September 1990. The only serious casualty was a Mexican robot that arrived in a box marked "fragile" but sadly looked like it had been dropped from a great height somewhere between South America and Europe.

Unlike a normal Olympics, where there are a number of preset competitions with preset rules, the route taken by the organisers was a little different. Instead, anyone with an interesting robot with an interesting set of behaviours was invited to attend and, upon registration, to indicate all the claimed capabilities of the robot. The organisers then simply looked through the list and, whenever any common capabilities were shown, a competition was devised. In this way eleven Olympic events were selected, namely wall-climbing, two legged races, more than two legged races, javelin, talk- ing, obstacle avoidance, phototrophic, pole-balancing, manipulation, behaviours and wall-following.

The reason for this apparent back-to-front way of organising the events was in response to some of the problems with existing international robot competitions. Two of the best known are the micro-mouse competition and the world robot table-tennis championships. In both of these competitions the rules are such that they focus the robot-builder into a very narrow niche which does nothing more than produce better maze-searching robot mice and better ping-pong playing robot arms. In short, they do little for advancing our understanding of general robot problems such as perception or learning. The purpose of the Olympics was more to show what is possible, what the state-of-the-art is, and to bring this new and fascinating technology to a broader audience. It also brought the whole event closer to the Olympic ideal of participation being more important than winning.

Whilst some of the robots were put together by amateurs, many were developed with serious applications in mind. This was particularly true for the wall-climbing competition, where the British and Soviet entries were already prototypes for application work in the nuclear industry. Furthermore, this event clearly showed how some novel idea could easily lead to a quantum leap in this technology. All the wall climbers used feet with suction pads and all used the idea of some feet holding on whilst others slid forward on legs or rails. The winner, however, was an order of magnitude faster than the rest and used a quite different approach. The robot, called ZIG-ZAG, was formed like a parallelogram with a suction cup on each corner. By extending and contracting the parallelogram with the appropriate application and release of the suction cups, the small machine shot up the wall in record time.

The pole-balancing event was also keenly fought with technologies close to industrial applications. Many control problems such as those associated with rocketry, satellite dynamics or the process industries can easily be decomposed into problems similar to those associated with pole-balancing. With entries from Hewlett Packard, Salford University, Turing, Lancaster University and the UK National Engineering Laboratory, the competition was underway. The entry from Turing was especially interesting in that it made use of a machine learning algorithm which learned for itself how to balance a pole. However, the eventual winner was the Salford device which put the others to shame by being able to balance one pole on top of another pole impressive stuff.

The biped event proceeded at a leisurely pace with the eventual winner from Cardiff in Wales completing the 10-metre race track in around 8 minutes. Things were a good deal faster in the multi-legged race when Penelope from Edinburgh University came thundering home well ahead of the American entry, Ghengis from MIT. It all went to show that when it comes to mechanical spider racing, eight legs are faster than six.

Live Action from the Robot Olympics involving two wall-climbing robots. The nearer of the two is Robug from Portsmouth Poly which is slightly in front of RVP2 from the Institute of Mechanical Problem Solving in Moscow. Robug later moved out of its lane, tried to clamber on top of the Russian robot and was disqualified!

In the Javelin, Gold went to a 140-year-old Victorian mechanical archer from the Museum of Automata in York, which soundly put to shame many of the high-tech entries from Europe, whilst Richard the robot head won the speaking competition. Sadly, one Japanese talker had problems with pronounciation but made up for things by winning the wall-following. The most hotly-contested competition was the obstacle avoidance with twelve entries trying hard not to bump into a whole variety of objects designed to catch the unwary ultrasonic sensor or laser range finder. The gold medal went to Asterix from the University of Toronto, Canada (the only robot not to hit something) with Oscarfrom Edinburgh University taking the silver and Yamabico from Tsukuba University in Japan taking the bronze. Gold in the manipulation event went to the Belgrade/USC hand from Yugoslavia, whilst Alpha Photon from Kent University showed great prowess in following lights in the phototrophic competition. Other robots deserving special mention were a robot vacuum cleaner built by the local Inverkeithing primary school and a large anthropomorphic robot from India which checked the weather (important in Scotland) and counted the people it met.

In addition to robot winners, each of which received their medals on the winners' podium to the accompaniment of a national anthem, a single overall champion was selected by the team of judges. The champion was chosen on the basis of three things. First, the quality of the mechanical and electrical hardware build. Second, the sophistication of the resulting behaviour and third, for novelty. On these criteria, the Olympic champion was Yamabico from the University of Tsukuba which made a small speech in Japanese to thank the humans for inviting him to Scotland.

The event was sponsored by the National Westminster Bank, the Scottish Development Agency and the American IEEE. The event is to be held every two years, returning to Glasgow every fourth year. No decision has yet been made concerning the venue for the 2nd International Robot Olympics.

Peter Mowforth
The Turing Institute
Glasgow, UK