The European Commission has entrusted the technology centre Eurecat with leading the development of the first major European artificial intelligence model for robotics in the industry. The initiative will receive more than €40 million in funding and aims to help strengthen Europe’s industrial sovereignty.
This foundation model, to be developed within the European GRAIL consortium coordinated by Eurecat and presented today in Barcelona, “will be trained with large volumes of data from different environments, tasks and types of physical interaction. This will provide a common basis for robots to understand how to interact with the physical world in a reliable and adaptable way,” explains Néstor García, co-coordinator of the European project and head of Eurecat’s Robotic Manipulation group.
According to Daniel Serrano, director of Robotics and Automation at Eurecat, the system “will serve as a technological foundation applicable to different types of robots, such as industrial arms, collaborative robots and humanoid platforms”.
This will make it possible “to use previously acquired knowledge and skills and adapt them to new tasks or situations, reducing preparation time and facilitating deployment across different industrial sectors, in order to maximise impact on the economy and society”.
Magí Dalmau, co-coordinator of GRAIL and head of Eurecat’s Cognitive Robotics group, says the creation of this major European model “will help move towards robots that can better understand their environment, plan their actions and carry out tasks more autonomously, always with mechanisms that allow people to supervise their behaviour and intervene when necessary”.
The initiative will therefore contribute “to strengthening European industrial sovereignty by promoting the development of Europe’s own solutions in artificial intelligence and advanced robotics,” says Simona Neri, project co-coordinator and Innovation Manager of Eurecat’s Robotics and Automation Unit. The aim, she adds, is for “European industry to have access to reliable and competitive technologies, while reducing dependence on external models and platforms”.
Technology validation in five strategic sectors
The technology will be validated in five key European sectors, including automotive, aerospace, intralogistics, steel and electronics, to ensure that the foundation model responds to the needs of European industries.
The tests will be carried out with different types of robots to demonstrate that the technology can work in diverse environments and applications.
The initiative plans to allocate €10 millions of its budget to expanding its impact through financial support for small and medium-sized enterprises, large corporations, research centres and other actors in the European ecosystem.
This support will be channelled through three Open Calls, which will fund more than 40 complementary projects focused on data training and the validation of the foundation model in other industrial sectors. Funded projects will receive support from project partners and may access four unique European infrastructures in industrial robotics and AI.
GRAIL, which stands for Generative Robotics & AI for EU Industrial Leadership, is funded by the European Union under the Horizon Europe programme, within the call aimed at developing robust and reliable generative artificial intelligence for robotics and industrial automation.
Consortium
The consortium brings together around twenty European partners from 10 countries. These include five technology and research centres: Eurecat, the German Aerospace Center, the Centre for Research and Technology Hellas, AIMEN Technology Center and the Barcelona Supercomputing Center (BSC); six universities: the Imperial College London, the Delft University of Technology, the Technical University of Darmstadt, ETH Zurich, the Technical University of Nuremberg and the University of Patras; and ten companies: Neura Robotics and Comau in the robotics industry, as well as FundingBox, Aleph Alpha, Siemens Industry Software, Engineering, Diehl Aviation, Signify, Steinbeis 2i and Multiverse Computing.
