Ing. Libor Pacovský, MSc., BA

2025 - 2027

The ACDRC Embedded AI Processor for Automotive Applications project aims to intelligent driving technologies enhance driving safety through sensing, analysis, and decision-making. Central to these technologies is the ability to perceive the surrounding environment using sensors, which collect data and process it using embedded AI processors. These processors provide efficient AI computation, critical for intelligent driving systems machine learning (ML) and artificial intelligence (AI) techniques. Current embedded AI processors are often designed for general purposes, which may not meet the stringent requirements of automotive applications, such as low power consumption, high efficiency, and high performance. The specific needs of automotive intelligent driving applications demand optimized solutions. This project aims to develop an FPGA-based embedded AI processor optimized for specific automotive intelligent driving applications. The processor will handle 1Mp30 image signals, outputting detected object boundary coordinates and object categories.

2026 - 2028

The project aims to develop and implement a joint Czech-Austrian (distributed) test infrastructure enabling optimization and scheduling of tasks in data centers or other local computing clusters within three years to achieve high energy efficiency and sustainability by enabling the use of renewable energy sources.

2026 - 2028

Projekt THOR přináší inovativní řešení pro detekci, analýzu a interpretaci kybernetických hrozeb na základě doménových jmen v šifrovaném provozu. Využívá pokročilé strojové učení, neurosymbolickou AI a federativní učení s cílem posílit ochranu kritické infrastruktury a zvýšit efektivitu práce bezpečnostních týmů, zejména na pracovištích typu Security Operations Center (SOC). Řešení umožní včasné odhalení i dosud neznámých hrozeb a automatizovanou tvorbu Threat Intelligence, čímž přispěje k ochraně kyberprostoru. Bude také přímo nasazeno v síti správce české domény, CZ.NIC.

2026 - 2028

Projekt si klade za cíl výrazně posílit možnosti využívání biologických dat v oblasti umělé inteligence (AI) a řešit klíčové výzvy v personalizované medicíně a veřejném zdravotnictví. Prostřednictvím strategické spolupráce mezi českým a švýcarským uzlem sítě ELIXIR vznikne robustní platforma propojující molekulární, metabolická a klinická data s nejmodernějšími technologiemi AI. Zvláštní důraz bude kladen na vývoj metod a nástrojů pro interoperabilitu dat, bezpečné sdílení citlivých informací a využití AI v klinické a biomedicínské praxi. Úspěšná realizace projektu posílí vedoucí postavení České republiky a Švýcarska v evropském výzkumném prostoru, rozšíří kapacity obou zemí pro tvorbu a sdílení špičkových datových zdrojů a technologií a nabídne inovativní řešení, která budou přínosem pro tisíce uživatelů v oblasti biomedicíny.

2026 - 2029

ISIDORe II (Integrated Services for Infectious Disease Outbreak Research II) represents the strategic continuation of ISIDORe. Building on its proven capacity to mobilise Europe’s research infrastructures (RIs) for pandemic preparedness and response (PP&R) research, ISIDORe II consolidates the model into a leaner, strategically aligned consortium that enhances agility, coherence, and sustainability. Infectious disease–focused RIs form the operational backbone, ensuring depth in epidemic- and pandemic-relevant science, while transversal RIs provide enabling technologies such as imaging, omics, and structural biology. This integrated architecture ensures that Europe’s RI landscape delivers high-quality, challenge-driven, and interoperable services. ISIDORe II pursues three overarching missions: (1) deliver excellent RI services for PP&R research and empower users through a curated service catalogue, an interoperable digital platform, and inclusive outreach; (2) amplify impact by enabling scientific excellence, innovation, and data stewardship, accelerating medical countermeasure (MCM) development, embedding FAIR principles, and aligning with European Open Science policies; and (3) anchor the ISIDORe model within the EU and global preparedness ecosystem through coordination with actors such as HERA, ECDC, WHO, and DURABLE, while developing a stakeholder-validated sustainability plan. By offering a coherent, strategically sharpened model for transnational and virtual access (TNA/VA), ISIDORe II ensures that users benefit from comprehensive support, including ethics, biosafety, and regulatory guidance, alongside interoperable data practices. Its activities generate scientific excellence, policy-relevant evidence, and long-term impact for European and global health security. As the natural continuation of ISIDORe, ISIDORe II transforms a successful pilot into a sustainable cornerstone of the European preparedness and response landscape.

2026 - 2028

Pathfinding in graphs is one of the most fundamental problems in graph theory. Dijkstra’s algorithm for finding shortest paths between nodes has been a part of standard computer science curriculum for decades. However, in several applications in logistics and operations research, the desired path should satisfy some additional constraints, and thus cannot be found by a straight-forward application of Dijkstra’s algorithm. For example, on some crossroads turning right is not permitted, i.e. certain edges of the graph cannot appear as consecutive edges on the path. Another example occurs in rideshare route planning, where each pick-up location needs to be visited before the corresponding destination. In other words, the ordering of the vertices on the path needs to respect a given partial order. Yet another case is travel planning: it is often convenient to choose connecting flights that belong to the same airline alliance. This can be modeled as an edge-colored weighted graph, where we are searching for the shortest monochromatic path. The three above scenarios might seem very different, but they share one key property that makes pathfinding difficult. Namely, Dijkstra’s algorithm relies on triangle inequality, which does not hold in the above cases. The aim of this project is to design novel algorithms that overcome this obstacle and to conduct a systematic study on different variants of the pathfinding problem with traversal constraints. To achieve this, we will employ parameterized complexity tools, whichenable us to study the complexity of problems with respect to multiple parameters describing theinput and output data.