Strengthening strengths - the world's first "joint competition" of the most advanced robots in Europe



Humanoid robots use artificial intelligence to collaboratively solve tasks that address some of the major challenges facing society, such as demographic change. The world's first Robotic Coopetition in Nancy, France, brought together 150 leading robotics experts from all over Europe and 20 robots from 25 to 28 November 2024. The European network of excellence for AI-based robotics is coordinated by the Institute of Robotics and Mechatronics. Their common mission: a coopetition, i.e. the combination of collaborative work (cooperation) and competition, to inspire new developments in robotics. The teams work on the most advanced robotic systems worldwide, such as humanoids, industrial robots, robotic centaurs or quadrupeds on wheels. The four-day event is part of the euROBIN project, which is coordinated by the German Aerospace Center (DLR). euROBIN is the European network of excellence for AI-based robotics, bringing together top research centres and major European industrial partners.
Coopetition is a new competition concept designed to promote collaboration and exchange. Not only are the teams' technological solutions assessed, but also their flexibility in integrating functional software modules and data from other teams into their own system. The aim is also to provide algorithms that are as high-quality as possible so that the other teams are happy to use them in turn. "The excellence partners share their results on our platform - for example, data such as environment maps, object models, execution plans and algorithms for perception, decision-making and cognition," explains Prof Alin Albu-Schäffer, euROBIN coordinator and Director of the DLR Institute of Robotics and Mechatronics. "The researchers select the best and most useful solution. They try to contribute the most flexible and fastest way in given application scenarios. In this way, small teams in particular can benefit from the experience and cutting-edge research of others and concentrate on their own expertise. 'Strengthening strengths' is the network's motto so that we can progress as a community as a whole."
The open exchange shows, among other things, what makes software or certain data sets flexible to use and therefore particularly successful for the market. It is still extremely difficult and costly to equip a robot system with "foreign" technologies and capabilities, as the components have to be harmonised and specially reprogrammed.
The latest methods of artificial intelligence (AI), such as the large language models of ChatGPT and others, are now opening up completely new possibilities that are being researched in coopetition. The transferability of robotic approaches is an important piece of the puzzle for their further development - because robots must learn to find their way in a wide variety of environments in order to be of use to people where they are really needed. The euROBIN network therefore aims to stimulate new solutions that enable the joint development of cutting-edge technologies and continue to strengthen Europe's leading role in robotics.
Common goal: robotic assistance in everyday life
A total of 22 teams took part in the challenge at the Prouvé Congress Centre, divided into the categories of industrial robots, outdoor robots and assistance robots. The DLR Institute of Robotics and Mechatronics competed there with its humanoid robot Rollin' Justin. For the assistance robot league, a household environment with different kitchens was set up in which Rollin' Justin had to perform various tasks as autonomously as possible with the help of its artificial intelligence when called upon.
These tasks are modelled on everyday assistance tasks: The assistant robots were given randomly selected and previously unknown assistance tasks in the kitchen by the referee. Rollin' Justin had to find and pick up objects and place them on a table or work surface, open drawers or a dishwasher, or hand the objects over to a human - and all this in different kitchens. In the spirit of euROBIN, special points were awarded for actions in which a robot used the software or previously provided data from a project partner.
Robots will soon be able to act in every household without having been specially programmed to do so. Even if this seems quite simple for humans, a robot needs to know a great deal about the respective environment and objects. Or it must have a high degree of transferability, which is made possible by AI. One challenge is that components that are not part of the system have to work together: If a robot rolls on four wheels, for example, how can you transfer what it has learnt to a two-legged robot? As the software cannot be adopted one-to-one and can only be adapted approximately, inaccuracies arise when performing the task. These inaccuracies add up the more external components are integrated and the more complex they are.
Pioneering role in AI-based robotics
euROBIN is Europe's answer to the international challenges in AI-based robotics, which is developing rapidly and promises enormous growth potential. The project, which is funded by the European Union, brings together 31 renowned research institutions and companies from 14 countries. As the coordinator of the euROBIN network, the DLR is driving forward cooperation between science and industry. The aim is for Europe to make the next technological leap in robotics and further strengthen its pioneering role in robotics.
Cooperation between robots
It is impossible to imagine industrial production without robotics. In future, it will also play an increasingly supportive role in everyday life, for example in the form of humanoid robots that help people in the home and with care. euROBIN has created new momentum with the world's first coopetition. This is because one of the realisations of recent years is that the mere exchange of knowledge is not enough - research institutions and companies must also work on problems together.
"Germany and Europe are leading the way in the development of robotic assistance technology," emphasises Prof. Albu-Schäffer. "Co-bot technology, i.e. robots that can work directly with humans, is based on our European developments. We must now use this lead to face the major social challenges such as de-industrialisation, environmental protection and demographic change and not only keep up with the competition outside Europe, but also continue to shape it together. Through euROBIN, we have created a new, sustainable platform, established new ways of cooperation between the top European institutes and industrial partners and thus jointly developed solutions and new approaches," continued the coordinator.
As a research participant, DLR uses the raw data obtained during the coopetition to create training data for AI learning programmes and share it with the network partners. In future, robots will not only be able to solve tasks independently, but also learn from other robots. Different robots will then be able to complete a task together and complement each other with their respective strengths.