Moteur de recherche d’entreprises européennes
Financement de l’UE (14,4 M €) : Boîte à outils ouverte de perception, d'apprentissage et de prise de décision pour l'IA robotique à usage général Hor 11/09/2026 Programme de recherche et d'innovation de l'UE « Horizon »
Vue d’ensemble
Texte
Boîte à outils ouverte de perception, d'apprentissage et de prise de décision pour l'IA robotique à usage général
Most current AI systems rely almost exclusively on fast, intuitive System 1 processing, that is pattern recognition from large datasets, while lacking the deliberate System 2 reasoning needed for planning, adaptation, and trustworthy decision making. This imbalance limits adaptability, safety, and explainability in dynamic, real world environments. The OPERA project addresses these gaps by advancing General Purpose AI for robotics that integrates System 1 efficiency with System 2 analytical capabilities, both in inference and learning. This enables robots to estimate confidence levels and detect anomalies while maintaining fast reactive behaviour, escalate to System 2 when needed, plan over long horizons, adapt to domain shifts, and make explainable and trustworthy decisions aligned with European values and the AI Act. To achieve this, OPERA will combine hybrid neuro symbolic and data driven reasoning, continual and evolutionary learning for adaptability, GenAI based world models for safe training and digital twins, and the integration of reinforcement learning with self-supervised learning for explainable navigation and manipulation. It will further embed System 2 driven learning, active exploration, relational and physics-based reasoning to enhance efficiency and safety. The resulting modular, open GPAI toolbox will be validated in three use cases: smart manufacturing, advanced robotics, and maritime mobility, and fully integrated into the AI on Demand platform. By delivering insights into learning synergies and cognition inspired architectures, OPERA will provide validated System 1 and 2 tools that push GPAI beyond pattern recognition. OPERA will position Europe at the forefront of GPAI-enabled robotics, driving productivity, sustainability, and climate neutrality through energy-efficient, adaptive, and trustworthy systems, and achieving TRL 5 in the targeted use cases achieving open strategic autonomy in GPAI, the most significant emerging enabling technology.
| Aarhus Universitet | 754 710 € |
| Albert-Ludwigs-Universitaet Freiburg | 1 221 073 € |
| Aristotelio Panepistimio Thessalonikis | 2 484 500 € |
| Ceske Vysoke Uceni Technicke V Praze | 766 000 € |
| Consiglio Nazionale Delle Ricerche | 838 963 € |
| CY.R.I.C Cyprus Research and Innovation Center Ltd. | 707 500 € |
| Cyberbotics Sàrl | 929 750 € |
| Norges Teknisk-Naturvitenskapelige Universitet Ntnu | 827 353 € |
| Pal Robotics SL | 905 000 € |
| Robert Bosch Car Multimedia GmbH | 530 254 € |
| Robert Bosch Manufacturing Solutions GmbH | 325 929 € |
| Tampereen Korkeakoulusaatio sr | 1 665 000 € |
| Technische Universitat Nurnberg | 965 625 € |
| Technische Universiteit Delft | 1 432 926 € |
https://cordis.europa.eu/project/id/101298363
Cette annonce se réfère à une date antérieure et ne reflète pas nécessairement l’état actuel.
