RoboCup: not just fun and games

Home Q&A RoboCup: not just fun and games

Hot on the heels of the robot soccer World Cup in Brazil, robotics expert and RoboCup world champion Gerrit Naus gives the lowdown on everything from the tricks and technology behind the game to the latest robotic innovations set to transform our daily lives.

A humanoid robot

Gerrit Naus from Medical Robotic Technologies B.V. is all about robotics. From designing a control system for welding robots to founding a company specialising in medical robotics, the young Dutch innovator is dedicated to developing high-tech solutions that will make people’s lives easier. But he also likes to have some fun while doing this.

That’s why, for the past six years, he has been one of the brains behind his university’s robot soccer team, Tech United of Eindhoven University of Technology, which won the Middle Size League of this year’s RoboCup.

TECHNOLOGIST Congratulations on Tech United’s RoboCup victory! What made your robots better than the opposition’s?

GERRIT NAUS Thank you. The main advantage of our players is their cooperation. They truly play soccer as a team. Other teams are strong in smart one-on-one play and accurate shots. Our biggest challengers are the teams from China, Portugal and Iran. Both China and Portugal are very accurate with their long-distance shots. Iran and China have very fast and sturdy players with a lot of power that have strong one-on-one play, but also often prevent playing soccer by bumping into the opponents. We rely on smart positioning on the field enabling passing between our players, which we combine and alternate with long-distance shots.

In our view, passing is the way to progress the game in future, so we have been investing in the underlying techniques for several years now. If the passing is a bit slow or not accurate enough, it’s very difficult to play against the fast robots of Iran and China. But this year we managed to optimise the technique far enough by maximising the speed of passing and shooting, which resulted in the victory. We’ve also put a lot of effort into our goalkeeper, enabling it to catch long-distance shots. The difficulty with these shots is that the keeper has very little time to react and must rely on an estimate of a flying ball. This estimate is based on information from several cameras, including two Microsoft Kinects that determine the exact location of the ball entering the goal. Summarising, I would say we won with ‘totaalvoetbal’.

TECHNOLOGIST But RoboCup is not just fun and games. What are some of the ‘serious’ benefits you aim to get out of developing world champion soccer robots?

GERRIT NAUS The soccer game is just the platform chosen by the RoboCup organisation as a very challenging environment. It takes years of practicing and a lot of talent for a person to become a world-class soccer player. On the other hand, everybody knows the game and for many it’s a lot of fun to play. So soccer is chosen as a platform to catalyse the development of robotics in general.

A humanoid robot picking up an apple

Photo: Bart van Overbeeke

Robotics is all around us, from our everyday cars to autonomous Mars rovers. We need robotic assistance to stretch the limits to the extent that we want them to; think of production speed, comfort of driving and living, working in hazardous environments, precision of tasks, for example in surgery – and the list goes on.

Currently, the focus is still often on integration, which makes the RoboCup Middle Size League such a nice league to participate in. In the future, when platforms become more standardised and affordable, the focus will gradually shift to artificial intelligence. This is a development we already see in the RoboCup leagues, but it’s often limited by the platform. The benefits of a standardised and affordable platform are enormous – just consider the implementation of robotics in the household, in rescue operations, in daily life!

TECHNOLOGIST What are the next big challenges within your field of robotics?

GERRIT NAUS Human acceptance of robotics is a big challenge. Understanding that every car nowadays is a robot and understanding the true benefits of it is difficult. People often envision a robot as a humanoid system that walks around and copies human behaviour. I believe that we should focus on the benefits of robotics rather than their ‘coolness’. We should get rid of the idea that a robot per definition equals a humanoid system. Certainly in the short-term, the implementation of robotics in the household will not result in a humanoid system that copies human behaviour, but more likely a Middle Size League robot that can bring you coffee. Think of the robots in the @Home leagues, which, effectively, are soccer robots with arms.

Gerrit Naus with a robotic assisting system for surgery.

Gerrit Naus’ company is currently commercialising PRECEYES, a robotic assistant for eye surgery.

Realising that a robot can be much more (actually less) than a humanoid, and that we are already surrounded by robotics in our daily life today, will help us accept the term ‘robot’ as a positive thing. For instance, surgery aided by a robot means that a robotic instrument can filter away tremors of a surgeon so he or she can perform surgery with higher precision. We have to think of the benefits robotic assistance can provide and exploit those where needed.

A second big challenge is collaboration between robotics and humans. Who (or which) is right? In case of autonomously operating systems such as the soccer robots, this is a relatively small issue; they play soccer on their own and after half an hour you know whether they won or lost.

In case of automating or supporting human tasks, cognitive collaboration becomes an important issue. In robotic surgery, for example, questions might arise if the robotic instrument detects that the surgeon is about to penetrate a delicate tissue: is this on purpose or not? If yes, to what extent should the robot allow the surgeon to do this? Or should it completely ignore either its measurement or the surgeon?

With Medical Robotic Technologies, an Eindhoven University of Technology spinoff, we target robotic assisting systems for surgery. These systems are not autonomous and merely assist the surgeon in performing surgery at or over the limit of what’s possible to do by hand. This way, we fulfil an unmet need, enabling surgeons to develop treatments for diseases they cannot currently treat by hand. This doesn’t mean robotics take over surgery – it means robotics enable surgery.

TECHNOLOGIST Now, let’s talk a bit more about robotics within the context of RoboCup… 

GERRIT NAUS RoboCup uses the soccer game as a platform to further develop robotic technology. The goal is to beat the human world champion soccer team in 2050. So every year, a game is organised between robots and humans at the end of the RoboCup tournament.

Tech United’s Middle Size League soccer robots.

Tech United’s Middle Size League soccer robots in action at RoboCup. Photo: Bart van Overbeeke

Traditionally, the robots of the Middle Size League are invited, as their game play resembles the human soccer game the most – today. The RoboCup organisation envisions that humanoid robots will play against humans in 2050. But for now, the biggest challenges with humanoid robots are not yet on the level of game play. The first goal is to make a robust and reliable walking robot.

Besides the Middle Size League, there’re for instance the Simulation League, the Small Size League, the Standard Platform League and related @Home and Rescue leagues. Considering soccer, the Middle Size League brings together all the difficulties of the other soccer leagues. For instance, the Simulation League misses out on the complexity of hardware, and the Small Size League robots rely on a field-side camera system instead of being truly autonomous. So I believe that at the moment, the Middle Size League is by far the main soccer league of RoboCup – and therefore also the most fun. (Here you can watch a goal that our robots scored against the human team this year.)

TECHNOLOGIST Could the other RoboCup league winners claim their league as the main one?

GERRIT NAUS As I mentioned, you could claim that the Humanoid League is the main one. However, copying a human might not be the smartest way to develop a robot. Walking is difficult, while driving is often relatively easy. On the other hand, this difficulty makes it especially challenging, stimulating research and development in robotics. That’s why the RoboCup organisation has formulated their goal. The Middle Size League robots are small carts, but that might be a way more suitable and feasible solution than a humanoid as a robot for future applications. Therefore, as a help in the household, the first commercial robots will definitely be driving instead of walking.

TECHNOLOGIST Do you have some tricks up your sleeve for raising your game next year?

GERRIT NAUS RoboCup is an open-source community in which all information and developments are shared. You can download the drawings of our robots and all software for free and build your own robot team. The difficulty in the Middle Size League is the integration of all parts: software, hardware, wireless communication, electronics, camera systems, and so on. Every new component introduces difficulties in robustness, reliability and functionality the first time it’s introduced. So, although development happens rapidly thanks to the open-source nature of RoboCup, maximising the benefit of an innovation can take years.

The passing between robots is a good example. One of our goals is to exploit this further next year through better positioning and more reproducible shots. We’ve also been working on a new platform that’s sturdier and faster, comparable to the Iranian and Chinese robots. A prototype has already participated in some of this year’s matches – with good results. We might introduce it at large next year.


A new miniature robot developed by EPFL researchers can swim with fish, learn how they communicate with each other and…