EU Launches REMARO Trusted AI Research Project for Underwater Robotics, Trains 15 Researchers

The REMARO project, which aims to develop trusted AI for underwater robotics, is being carried out as part of the EU’s Horizon 2020 research and innovation framework program. Details are available in a CORDIS fact sheet published as of February 24, 2026. REMARO aims to train 15 early-stage researchers (ESRs) and develop the first methods to quantify the reliability of underwater robotics AI.

REMARO represents a European effort to ensure AI reliability in underwater robotics and could contribute to improving environmental monitoring and the safety of industrial infrastructure. It is also an important project that could shape the direction of technology development by presenting a new approach to AI safety research.

The REMARO project is novel in its focus on AI reliability in underwater robotics. While conventional AI research has focused on improving performance, this project seeks to introduce “quantified reliability” to underwater robotics AI for the first time, positioning it within the field of AI safety research. In particular, the principle of designing safety and reliability alongside perception points to a potential paradigm shift in AI system development. The project combines researcher training (15 people) with a concrete deliverable (the first trusted AI methods), reflecting the characteristics of the EU’s Marie Skłodowska-Curie Actions research funding program. The figures—about 40 days of activities, three challenge camps and 37 secondments, including 20 at industrial research laboratories—indicate a practical training program that emphasizes collaboration between industry and academia. In terms of implications for the industry, underwater robotics plays an important role in a range of fields, including wind energy, oil and gas, food, pharmaceuticals and mining. Automating monitoring processes can help reconcile environmental protection with economic growth. The trusted AI developed by REMARO could directly contribute to preventing incidents in these industries, including power outages, oil spills and pollution. However, it will take time for the project’s results to be applied in industry. Training researchers and developing methods are at the basic research stage, and further testing and demonstrations will be needed for practical use. Open licenses for software and data could benefit the research community, but challenges remain from a commercialization perspective. Open questions include the specific performance metrics for the AI methods under development and their adoption by industry. The extent to which the project’s results will affect Europe’s underwater robotics industry will depend on future demonstrations and market response.

REMARO views the underwater robotics industry as an important part of the European economy and aims to automate monitoring processes in areas including wind energy, oil and gas, food, pharmaceuticals, mining, and the biological and Earth sciences. The project brings together experts in robotics, AI, software reliability and safety, and will train 15 researchers. The AI methods it develops will be the first for underwater robotics to include quantified reliability, accuracy specifications, models, testing, and analysis and verification methods. REMARO is based on two fundamental principles: 1. Underwater robot autonomy requires a comprehensive hybrid deliberative architecture; 2. Safety and reliability must be designed alongside perception, not added afterward. The program includes about 40 days of intensive activities, three cross-sector, cross-disciplinary challenge camps, and 37 secondments, including 20 at industrial research laboratories. Training materials will be published in the REMARO Book and online learning hub, and software and data will be licensed for open use.