INESC-ID gets new projects approved in three different FCT calls
INESC‑ID recently saw four of its proposals approved by Fundação para a Ciência e a Tecnologia (FCT), across three competitive calls that reinforce the institute’s role in national and international research initiatives.
The Defence + Science Programme call focuses on strengthening scientific and technological capacity in areas of strategic relevance for National Defence. It promotes closer collaboration between research institutions and the Armed Forces, supporting multidisciplinary work on societal challenges and security, and contributing to the broader development of the defence economy.
The Call for Exploratory Research Projects under the CMU Portugal Program aims to support high‑impact exploratory research developed by Portuguese institutions, Carnegie Mellon University, and industry partners. Centred on Information and Communications Technologies, it encourages projects that use the data economy to drive innovation and growth, led by recent PhD graduates and researchers affiliated with Portuguese research organisations.
Finally, the Call for Exploratory Research Projects under the UT Austin Portugal Program 2025 aims to fund high‑risk, high‑reward collaborations between Portuguese institutions and UT Austin. By reinforcing national and international competitiveness in science and technology and promoting innovation and knowledge transfer, it invites proposals with clear outcomes and strong potential for future development in priority areas such as advanced computing, clean energy, nanotechnologies, and space‑earth technologies.
The approved projects and their respective calls are as follows:
DEFENSE + SCIENCE: Call for Exploratory Projects 2025
Establishment of Secure Communications Using Dual Orchestration: QKD and PQC
PI: Manuel Goulão
The project responds to the growing risk that some future quantum computers will be able to break today’s cryptographic systems and compromise long‑term sensitive communications. To counter this, it proposes the development of a hybrid key‑establishment system that blends Quantum Key Distribution (QKD) with post‑quantum cryptography (PQC), creating a layered defence in which communications remain protected as long as at least one of the component schemes remains unbroken, providing robust protection. In doing so, the work also reinforces and contributes to the broader European effort underway in the ORQESTRA project.
LLMs for Automated Configuration of Command and Control Systems
Modern military operations rely on multiple Command and Control (C2) systems that must be configured in a consistent and reliable way, yet this remains a manual and error‑prone process that can undermine interoperability and mission safety. The project tackles this challenge by creating an LLM‑based framework, LLM2C2, capable of turning natural‑language instructions into compliant configurations that are automatically checked against formal logical rules until all constraints are satisfied.
Call for Exploratory Research Projects under the CMU Portugal Program 2025
Quantitative Program Analysis for Improving LLM Code Generation
LLMs are increasingly used for coding, yet they still produce subtle bugs, misinterpret program behaviour, and react unpredictably to small prompt variations, issues that can lead to serious security vulnerabilities in real software‑engineering contexts. This project aims to strengthen their reliability by integrating program analysis directly into the training and decoding pipeline through a new approach, “Reinforcement Learning from Static Analysis Feedback”, which guides models toward semantically correct, specification‑respecting code. By moving beyond today’s syntactic and test‑based evaluation criteria, the work has the potential to significantly improve the quality and trustworthiness of LLM‑generated software.
Call for Exploratory Research Projects under the UT Austin Portugal Program 2025
EXPLanation and ObseRvation for HumanOid DOmestic Robots
PI: Bruno Martins
Robotic systems need to turn complex sensory input into meaningful reasoning and interaction, yet current AI architectures still struggle with long‑horizon decisions and with collaborating naturally alongside humans. Building on the emerging strengths of LLMs in multimodal reasoning, explanation, and dialogue, the project proposes an LLM‑based framework that enables robots to reason in a grounded and explainable way while engaging in continuous, task‑aware spoken interaction during demanding domestic activities and collaboration with humans.