 Last edition of the FIRA Cup 2002 - a major robotic soccer competition - ended last week. The event, one of the two "Robot World Cups", was organized in Korea, and gathered more than 200 teams from 25 countries.
Competing in half a dozen categories with different tournaments and rules, the teams came from Universities and Research Labs in robotics and AI.
For the first time ever, FIRA Cup introduced a Humanoid Robot Tournament, called HuroSot. The biped robots had to compete in three tests, still far from an actual soccer game, but sufficient to demonstrate motion, dynamics or vision capabilities.
This is a "double" interview, with two researchers on humanoid robots that participated to the event.
Thomas Braunl is a Professor at the Dept. of Electrical and Electronic Engineering at the University of Western Australia, and was engaged in the FIRA Cup 2002, in the HuroSot category, while Dong-Han Kim, from the Korean Advanced Institute of Science and Technology, was the winner in this category with his team.
RobotsLife.com: From a general point of view, what were the main differences between the latest edition of the Robot World Cup and past editions?
Thomas Braunl: First of all, we should mention that there are currently two competing World Championships: FIRA (2002 in Korea) and RoboCup (2002 in Japan). The fact that the (human) Football Worldcup has a similar constellation is not a coincidence.
FIRA is the older and original competition, originally
limited to the "MiroSot" group of robots: size 6 cm x 6 cm x 6 cm and
radio-controlled with the help of an overhead camera. Today, there are
several sub-leagues, including even smaller robots (NanoSot) and the
pure simulation league (SimuroSot).
In FIRA World Cup 2002, there were two new leagues introduced:
RoboSot: Finally a small size league that requires robots with local
intelligence, on-board vision, sensing, and computation. This is one
of the exciting new developments for which we have pushed for a number
of years.
Because of the increased complexity, the level of play is still
inferior to the "remote controlled" other leagues, but this is much
closer to our research goals for autonomous intelligent robots. So watch out for this league in future years.
HuroSot: The Humanoid Robot Soccer League. FIRA was the first organization to introduce a humanoid league, with myself and Jacky Baltes from Univ. of Auckland, New Zealand as chairs. This year's three events:
robot dash (a sprint event), penalty kick (robots have to kick a ball
in an empty goal), and an obstacle avoidance parcour are to be seen
as preparation events for humanoid robot soccer in the following years.
We had a number of excellent robots in this competition, with two outstanding ones, both from Korea. From 2003 on, this competition requires robots to be fully autonomous, i.e. they will have to do their own image processing and sensing on-board.
RoboCup is the second world organization for robot soccer and has eveloved
faster in the last few years, therefore receiving much larger media
coverage. RoboCup has similar events to FIRA, including an overhead-vision
small-size robot league and a simulation league, but also a large-size
robot league without global sensors, the "Sony legged robot league",
and a few more.
Following FIRA, RobotCup has also introduced a biped league in 2002, but
still has no autonomous small-size league.
Dong-Han Kim: 1. This year’s FIRA Cup was held on a grand scale, with strong support from the Korea Science Foundation (KSF), Korea Robot Soccer Association (KRSA) and YTN.
2. A total of 207 teams from 25 countries, with over 600 professors, scientists, engineers and university students, participated in this grand event, making it the largest ever in its short history.
3. The event was organized into two stages. The preliminary stages of MiroSoT and SimuroSoT were held in the 5 FIFA World Cup cities, namely, Busan, Daegu, Daejeon, Gwangju, and Suwon, from 23 May 2002 to 24 May 2002. The finals of MiroSoT and SimuroSoT, together with the other categories such as HuroSot, were held at SETEC, Seoul, from 26 May 2002 to 28 May 2002.
4. HuroSot made its debut in this year’s FIRA Cup, and was an interesting highlight of the event.
5. In conjunction with the finals of 2002 FIRA Cup, the FIRA Intelligent Robot Exhibition also took place at the same venue, while the 2002 FIRA Robot World Congress took place at COEX, Seoul, from 26 May 2002 to 29 May 2002. The exhibition showcased the latest entertainment, industry and personal robots from over 30 Korean companies. The World Congress provided the opportunity for the sharing of new findings and knowledge by the international community, with particular emphasis on the development of robot soccer system in the subareas of image processing, communication, cooperation and other AI strategies.
RobotsLife.com: Concerning HuroSot, do you think this kind of initiative can play a major role, helping research on biped robots?
Thomas Braunl: I believe that this is definitely the case. I know for a fact that at least
two teams started research on biped robot developments, especially to
compete in the FIRA Worldcup, and I expect even more to follow over
the next year.
The effect of a competition like HuroSot in furthering research in the
area of humanoid robots cannot be stressed enough. The exchange of ideas
among the participating teams and also a generaly raised interesed in
this area will undoubtedly improve research efforts.
Dong-Han Kim: Definitely. In inventing the game of robot-soccer, an important goal of FIRA is to promote and stimulate the research and development of autonomous, intelligent and human-like robots. As a natural progression along the robot-soccer theme, the HuroSot category was introduced this year, and will be held annually.
In HuroSot, a robot player is more human-like in that it has two legs, hence the term humanoid. Given the current state of the art, this year’s participants were only expected to endow their humanoid robot with, for instance, the ability to walk steadily, avoid obstacles simulating stationary opponent players and take penalty shots, all under the remote guidance of its human trainer.
We believe that this game will become widely accepted as a benchmark problem to evaluate a humanoid (biped) robot integrated with the many multi-disciplinary technologies of dynamic control and balancing, sensor, vision and decision making, just to name a few. In subsequent competitions, the format of the HuroSot will develop in tandem with the state of the art developments in robotics. In this way, we believe the HuroSot initiative will stimulate research and development of humanoid (biped) robots. Although still a distant dream, it is hoped that the game will eventually evolve into one with two competing humanoid robot teams playing a full game of robot-soccer.
RobotsLife.com: In HuroSot, the three winners came from Asian countries. How can you explain that?
Thomas Braunl: FIRA is traditionally strong in Asia, but e.g. in the MiroSot (wheeled robot soccer) league, the runner-up was from Dortmund, Germany.
From my own experience, it is much much easier in Korea, Singapore or
Japan to get funding for a biped robot project than e.g. in Australia
or New Zealand. Many Asian countries highly value research efforts in
Robotics or IT in general and supply ample funds for reseach projects
in these areas. Unfortunately, this is not the case in Australia,
although there might be sufficient interest and expertise in the
research community.
For our own biped robot entry, we had very little funding for the project
and could only present a "budget version" of a biped walking robot.
Also, using too small feet proved to be too ambitious. In order to
perform better at the competition, we should have increased the support
area as allowed by the rules.
Dong-Han Kim: I think generally, those asian teams were better prepared than the others.
In fact, overall, Korea and China were the big winners of the 2002 FIRA Cup, having managed to clinch one of the top three places in all the major categories they participated in. Like Korea, China has been actively developing its robot-soccer games. In fact, prior to FIRA Cup 2002, the FIRA China Robot Soccer Association (CRSA) held its 3rd National Robotic Soccer Championship, from April 12 to15, 2002 at Daqing, China. More than 30 Chinese universities, including Beijing University, Zhejiang University, and the Harbin Institute of Technology participated in this championship.
The RIT (Robot Intelligence Technology) lab, that I am a member of, has been building up the expertise in the area of humanoid robotics for several years now. The HamSaRam (version III) humanoid represents an accumulation of our expertise in this area.. I think, compared to the other participant robots, HanSaRam III showed relatively higher mobility and dynamic stability, and this probably gave us a slight edge over our competitors.
RobotsLife.com: Would you say that a new kind of soccer, involving humanoid robots, has a bright future ? Do you foresee a future where such competitions become so popular that they are broadcasted worldwide on TV and supported by big sponsors in the same way the human World Cup is?
Thomas Braunl: At the moment I would not want to compare it with the human WorldCup,
but just have a look at the popularity of "Robot Wars". These destructive,
brainless (since remote-controlled) games have achieved high ratings
and a regular TV show.
Humanoid robots playing soccer - or competing in other sports events
like the onces we had at the FIRA WorldCup this year could become
similar if not more popular in the near future.
Dong-Han Kim: These days, a British company broadcasts a robot war and there are a lot of fan in Korea. I think that robot will step up to major programs in the near future.
RobotsLife.com: Do you think that the final aim of the RoboCup in Japan (a team of robots able to win against a human team, 50 years from now) is reachable?
Thomas Braunl: This goal has been mentioned by a few people from RoboCup and FIRA,
but I see this more as a media catch phrase. It is very hard to predict
what will be possible in 50 years time, since in the area of robotics
this involves so many different disciplines as in Mechanics, Materials,
Energy sources, Electronics, Sensors, Computer Hardware and Software.
I am convinced it will eventually be possible, but my estimate is it'll
take longer than 100 years from now.
Dong-Han Kim: Definitely. With the rate of the current advances in computer technology, artificial muscle and vision systems this may happen within the next 4 to 5 decades.
Interview by Cyril Fievet, June 5, 2002
Pictures: courtesy of Thomas Braunl and FIRA
Related websites
FIRA Cup
Thomas Braunl
KAIST
Previous interviews
Tadahiro KAWADA - Kawada Industries
All interviews
|
