|main site > home > Personal Robotics Seminar, London 1985 >||28 May 2007|
Contests such as Micromouse, Robot Pinq-Pong and Build-a-Robot can be viewed as a lot of fun. They have to be enjoyable to persuade the contestants to put in so many months of creative effort. But they have a more serious side.
In the six years since the announcement of Euromouse, great numbers of young (and not-so—young) engineers have come to grips with on-line computer control, sensor technology, stability theory and problem-solving algorithms. More important, they have succeeded in taking the essential elements of the theory and making them work in practice. Many of the entrants have no engineering qualifications, but engineers they are in spirit and achievement.
Robotics research is often fettered by the demands of industry. "You can't use this technique — it might not be cost effective. That one might upset the work force." No such limitations apply to the enthusiasts, and they incorporate novelties in their mice sometimes years befora industry sees them as valuable. Adaptation, learning and responsiveness to sensors are all essential to the Micromouse. Voice output was used several years ago; it enables the status of the mouse to be diagnosed.
Although Euromouse is a contest of speed, it can be taken gently. While exploring, the mouse can sit and think for a while without imperilling its score for its fastest run. That is not true of Robat the robot ping-pong contest. Here a moment's hesitation will miss the ball and lose a point. The level of vision coordination whicn Robat demands goes far beyond any industrial robotic task. But is it beyond the bounds of reason for robots to lob components to each other? A bricklaying robot might well be kept supplied by hod-carrier with an accurate serve!
The then Director of the Science and Engineering Research Council's Robotics Initiative pronounced that games such as ping—pong were clearly beyond the possibilities of robot technology. The contestants are obviously not sophisticated enough to realise this? they seem likely to succeed.
The history of Euromouse.
Euromicro '8O was to take place in London, and the conference organisers felt that some lighthearted attraction would leaven the serious technical papers. An account appeared in the American IEEE Spectrum magazine of a maze-soving contest with prizes for the fastest first run and the best learning run. It seemed a good idea, and the European contest was launched. Then the Spectrum accounts took on a sour note. A high speed 'dumb' wall—follower was outstripping all the brighter mice, and attempts were made to outlaw it - but was that really playing the game ? Something was clearly wrong with the rules.
The answer was of course to put the target at the centre of the maze. The paths could be highly connected, and by surrounding the centre with closed routes the wall-followers could be baffled for ever. By declaring the maze dimensions (16 by 16 for binary convenience) and the coordinates of the target, the emphasis could be placed on control, navigation, mapping and strategy. The mice have certainly excelled in all of these.
Two months before the London contest, a trial heat was held at the Open Day of Portsmouth Polytechnic. It taught everybody a lot. The first lesson was that mouse-builders are shy to show off their creations unless perfect. From two hundred applications, the number of contestants prepared to appear dwindled to two and they were both Polytechnic students! Much pleading by telephone the night before the event brought the numbers uo to just five.
Plessey's Fred' and Marconi's 'Meryl' were far from complete. Although they were both impressive in terms of their construction and concept, one could only spin in circles while the other was good-humouredly driven by manual switches to entertain the sizeable audience. Algernon's guidance circuitry was crossed, so that it could only run straight into the first wall; one of the Portsmouth mice had processor problems and bounced about at random. Only "Freewheelin' Franklin" made any real progress - and that was marred by a loose photocell connection. And yet the event was an enormous success. The audience seemed more delighted ay the disasters of the mice than by their successes. Among the spectators was Nick Smith; he had left his mouse at home.
Also in the audience were five delegates from the Japan Science Foundation, who took the rules back to Tokyo and built up their own contest to great heights.
The next two months were well spent, and the European finals at Euromicro 8O were quite a different matter. Mice arrived from all over Europe, "Midnight Sun" from Finland, "Lami" from Switzerland, "Superlite" from Germany and "Yamahico II" from Japan. "Fred" now took on a rodent shape, "Meryl" was under full control and a new mouse "Ancomical" was entered by ICL's amateur computer club. Tecnnology ranged from "Brainy Bricks", made from Lego, to "Pascal", sawn down from a toy car and struggling to clear the corners with three-point turns. Lami was marvellousiy engineered with tyres made of cross—mounted micro—wheels. These allowed it to perform a virtuoso display of driving in a circle whilst pointing North. The novel wheels unfortunately demanded an absolutely flat surface, and an uneven joint in the maze—base marred Lami's contest performance.
One mouse was built around a CMOS processor. The body of aluminium had been carved out with tin-snips, and the wall-sensors were metal flaps which closed contacts salvaged from a relay. With no previous electronic experience, Nick Smith had put together ''Sterling Mouse", the first mouse to reach the centre and "know" it had succeeded. The strategy had its roots in Dynamic Programming, but was simple enough in essence that the calculations could be carried out as a delay routine between motor steps.
An Easter workshop was organised by the ICL computer club, where ideas were exchanged and advances were made. Thezeus and Thumper appeared at Wembiey in 1981, founding a dynasty of winners. Thumper combined ingenuity with superb craftsmanship, with ingenious swivelling wheels which allowed it to manoeuvre without rotating. Thezeus established the practice of building a small personal computer into the micromouse - albeit with sawn-off keyboard. Fifteen mice took part in the Paris Euromicro finals, and Thumper became the new European champion. The rules had again been changed slightly, giving each mouse fifteen minutes in which to perform, the best run being counted. This put a premium on learning ability, and the time achieved for the best run had by now been cut to below a minute.
The 1982 British finals were held at the Computer Fair, Earl's Court, The first of three such years. Two new "Thezeii" beat Thumper into third place, but all three times were below one minute. The University of Tampere in Finland played host to the 1982 European finals; they snatched victory from Thezeus—4 by a mere two second margin in forty seconds, and retained the title the next year in Madrid.
At last in Copenhagen, at Euromicro 1984, Britain won the title back in the shape of "Enterprise", grandson of Thumper. The shortest path was seventy squares in length, covered in an amazing twenty-seven seconds. David Woodfield and Alan Dibley will now join teams from Finland and Germany in the Japanese "All World" contest in Tsukuba. Even more contestants are expected from South Korea and the United States, and of course Japan.
In November 1981, a micro-robotics conference at Imperial College provoked the question "what can follow Micromouse when solving the maze is seen as easy?" Three-dimensional mazes were suggested, along with non-cartesian shapes and walls which move. All these are possibilities, although there are still plenty of challenges in the contest as it is. A contest of a different type was needed, and the solution was robot ping-pong. Playing very safe, I named a date five years off for the first heat - but it is all happening in four.
A group of entrants met in Portsmouth on 19th January, 1985, to exchange ideas and polish up the rules. Three very primitive pieces of machinery arrived, two bat mechanisms and a vision system. Nothing really worked, although an oscilloscope trace showed a peak where tne ball might just be. Less than two months later, the contest was introduced on BBC's "Micro Live", One of the mechanisms now leaped about, threatening mayhem to the presenter who stood too close. The vision system put up an excellent screen display of the track of a real bouncing ball, and a completely new arm succeeded in taking a swipe at the ball.
At the European Personal Robotics Congress the Robats will do battle in earnest. I suspect that the flight of the ball will be erratic, to say the least. But from the first few tentative efforts, a whole new technique of dynamic robot interception will emerge.
In conclusion Micromouse has grown up here on a shoestring budget, begqing space at annual Brittsh exhibitions and scrounging prizes from the exhibitors. It is none the worse for that. In Japan, however. there is a permanent Micromouse Secretariat. They were present at last year's Euromicro Finals in Copenhagen, and awarded seven or more European participants free air flights to take part in this vear's Japanese finals. The maze used at the European Personal Robot Congress has been flown here by the Japan Science foundation, so that any incompatibilities can be sorted out beforehand, and mazes have been sent to South Korea and the United States. Could the importance which they obviously give to such contests be linked with their industrial success, perhaps in chicken-and-egg fashion ?
Micromouse, Robat and Build—a—Robot will continue to give qualified and unqualified engineers alike the opportunity to innovate, Their ideas may spin off into industrial applications, the contestants may themselves be recruited by marketers of new robotic products. In all events, they enhance the awareness and ability of the country as a whole to exploit the rushing tide of technology.