Objectives
High-Performance Computing (HPC) solutions used in mainstream systems like datacenters, handheld devices, and supercomputers, are becoming prevalent in embedded systems like those in automotive, avionics, space, railway, robotics, healthcare, industrial systems, and smart cities. However, embedded systems have specific sets of Non-Functional Requirements (NFRs) including safety, security, mission, quality-of-service, availability, reliability, power and temperature among others. NFRs shape how HPC solutions must be adapted/tailored for their adoption in embedded systems, and more generally, in systems where any of those NFRs can be regarded as critical.
The High-Performance Embedded Systems (HPES) laboratory aims at enabling the adoption of hardware, software, and artificial intelligent (AI) HPC solutions in embedded systems as its center of gravity, but also in any system with some form of criticality.
The HPES laboratory is the natural evolution of the former Computer Architecture – Operating System Interface (CAOS) group, born in 2005, whose constant growth and evolution led to this new organization in the form of a laboratory composed of the following three groups:
- HArdware Dependability for Embedded Systems (HADES). The HADES group has its center of gravity in the hardware support to enable the use of HPC devices and components in domains with NFRs such as safety, security, power, and real-time constraints among others, in relation to standards such as ISO 26262, DO254, and IEC 61508 among others. Activities in HADES include the proposal of hardware components for interference monitoring and containment, testing, diverse redundancy, timing predictability, acceleration, and support for low-level software layers such as hypervisors and GPU runtimes.
- SOftware Dependability for Embedded Systems (SODES). The SODES group investigates new tools and methodologies for software verification, with consideration of relevant safety standards such as ISO 26262, ISO 21448 (aka SOTIF), DO178C, and AMC 20-193 among others. SODES activities include system-level configuration of COTS MPSoCs (Multi-Processors System-on-Chip) providing appropriate support across the full software stack (hypervisor, operating system, middleware, and containers), dependable software architectures and patterns including AI-based software components, as well as the use of probabilistic methods and AI for the analysis of varying performance metrics.
- Trustworthy Artificial Intelligence for Embedded Systems (TAIES). The TAIES group focuses on the different dimensions of the use of Trustworthy AI in embedded systems and spans its field of application into systems with some form of criticality against the aforementioned NFRs. TAIES activities include risk modelling building on uncertainty modelling, causality and explainability for AI, tailoring AI solutions to meet all sorts of regulatory requirements such as legal and standard-related ones, and more generally considering unavoidable ethical aspects of the design and use of AI in embedded systems, in particular, and critical systems, in general.
By means of joint collaborations within and beyond the laboratory, the different groups aim at paving the way for the adoption of constantly evolving and increasingly powerful HPC solutions into systems with NFRs.
The HPES laboratory activities span from low Technology Readiness Level (TRL), where new concepts and ideas are conceived, to the highest TRLs where those concepts and ideas materialize into industrial-ready solutions transferred to the society and the market in the form of open source technologies, technologies licensed to existing companies, and the creation of BSC spinoffs where entrepreneurs from the laboratory create a joint startup with BSC to exploit technologies developed within the HPES laboratory. In this context, the track record of the laboratory (formerly CAOS group) backs this aim with the achievement of low TRL key projects, like ERC grants, and up to the creation of a spinoff and the transfer of technologies reaching certification in one of the most stringent application domains like the avionics one. In between, we have a wide variety of projects, either publicly funded (European, national, regional) or private contracts with industries across domains, but often in the avionics, automotive and space sectors.
HPES pursues to have societal and scientific impact while thriving for research excellence. The members of the HPES lab have a solid trajectory of participation in EU projects, national projects, projects with the European Space Agency (ESA), and bilateral contracts with industry. Team members also have a solid track record of international publications. Other merits include:
• In terms of technology transfer, group members have received several HiPEAC Technology Transfer Awards and created the Spin-off (Maspatechnologies S.L.) in technologies developed in the Laboratory and that was sold in 2022.
• Leadership of EU projects (projects led include PROARTIS, PROXIMA, SAFEXPLAIN, and METASAT) and leadership of ESA projects (projects lead include PROARTIS4SPACE, and GPU4SPACE).
• Personal grants including European Research Council (ERC) grants, Spanish Ramon y Cajal and Juan de la Cierva Grants, and Catalan Beatriu Pinos grants.
• Participation in a wide variety of EU programmes including FP7, H2020, Horizon Europe, ARTEMIS JU, ECSEL JU, KDT JU, and Chips JU.
• Best paper awards in conference of the area like DAC, ECRTS and ERTS.
• Conference organization such as ECRTS 2018, DFTS 2025, and ESWEEK 2026.
• Chairing and participation in key workgroups such as the RISC-V International Special Interest Group on Safety (vice-chaired 2022-2024), OpenHW Group – Safety & Security Task Group (co-chaired 2023-2024), DG CNECT -- RISC-V High Performance Automotive Platform Working Group (members 2022-2024).